#Wiki
1 messages · Page 1 of 1 (latest)
wiki in isolated mode is basically a separate curated knowledge vault.
Purpose:
- it gives you a structured, provenance-aware wiki
- it does not depend on
memory-core - it is good when you want the wiki to be its own source of truth, not a mirror of active memory
How to use it:
- enable
memory-wikiwithvaultMode: "isolated" openclaw wiki init- add material with
openclaw wiki ingest <file-or-url> - rebuild derived pages/indexes with
openclaw wiki compile - query with
openclaw wiki search/openclaw wiki get - make safe updates with
openclaw wiki apply - run
openclaw wiki lintafter meaningful changes
So the mental model is:
- isolated = “my wiki is its own maintained knowledge base”
- bridge = “build wiki pages from the active memory plugin’s exported artifacts”
If you want, I can also give you a tiny config example for isolated vs bridge.
How to build concepts In isolated mode, manually?
Yep, in isolated mode, concepts are basically manual Markdown pages in concepts/.
A good workflow is:
openclaw wiki init- ingest raw material into
sources/ - create a concept page in
concepts/<slug>.md - run
openclaw wiki compile - run
openclaw wiki lint
Minimal example:
---
pageType: concept
id: concept.rate-limiting
title: Rate Limiting
sourceIds:
- source.api-docs
claims:
- id: claim.rate-limiting.1
text: The API enforces per-user burst limits.
status: supported
confidence: 0.8
evidence:
- sourceId: source.api-docs
lines: 12-18
questions:
- Are limits different for batch endpoints?
---
# Rate Limiting
Short human-written explanation of the concept.
## Notes
- Applies to write-heavy workflows
- Might need retry/backoff
A few important bits:
pageType: concept- stable
id, usually likeconcept.something sourceIdspointing back tosources/...- optional
claims,questions,contradictions,confidence - after
compile, OpenClaw builds backlinks/related blocks/digests
So, manual concept building is really:
- sources = raw material
- concepts = distilled ideas/patterns/policies derived from those sources
Also, don’t hand-edit generated/managed blocks after compile, let the wiki own those.
Docs:
If you want, I can give you a best-practice concept template for isolated mode with source, claim, and synthesis examples side by side.
Ok
Nice 👍
If you want to keep going later, the simplest mental model is:
sources/= raw inputsconcepts/= distilled ideassyntheses/= higher-level summaries across concepts/sources
If this solved it, please right-click the solution message → Apps → ✅ Mark Solution so Answer Overflow can index it.
Ok but the agent do itself?
Not by itself, at least not just because isolated mode is enabled.
In isolated mode:
- automatic: compile/lint/backlinks/dashboards, and
autoCompileafter ingest if enabled - not automatic: inventing new concepts/entities from nowhere
So the agent can maintain it if you explicitly make that part of the workflow, for example:
- you ask it to ingest notes and create/update concept pages
- you give it wiki/file write tools
- or you run a cron/automation that tells it to curate the wiki regularly
If you want “pull from memory and organize it automatically”, that’s closer to bridge mode.
So short version:
isolated= curated by you or by an agent you instructbridge= imported from memory artifacts
If you want, I can sketch the best automation pattern for isolated mode.
How should be used concept in wiki they are read by the agent autonomously?
Yes, but not as “every concept page is always in the prompt.”
Best mental model:
What concepts are for
Use concepts/ for stable distilled knowledge, like:
- patterns
- policies
- mental models
- definitions
- architecture ideas
Not for:
- raw notes
- transient chat logs
- one-off tasks
How the agent reads them
Concept pages can reach the agent in 3 main ways:
-
Via wiki retrieval
wiki_searchwiki_get
-
Via shared memory retrieval
memory_search corpus=allmemory_get- this depends on the memory setup supporting shared corpus selection
-
Via compiled digest prompt
- enable
context.includeCompiledDigestPrompt: true - then a small compiled snapshot of top pages/claims is appended to memory prompt sections
- enable
So, autonomously?
Sometimes, yes, through retrieval.
Not usually as full-page automatic context on every turn.
So the right use is:
- put durable ideas in
concepts/ - attach
sourceIds/ claims / evidence - let retrieval pull them in when relevant
- enable compiled digest prompt if you want more proactive wiki awareness
If you want stronger autonomous use, I’d enable:
search.backend: "shared"search.corpus: "all"- optionally
context.includeCompiledDigestPrompt: true
Docs:
Bridge mode is working?
Yes, bridge mode is a real supported mode, but whether it works in your setup depends on the active memory plugin.
What bridge mode does:
- reads public memory artifacts
- can index dream reports
- daily notes
- memory root files
- memory event logs
Important catch:
- it only works if the active memory backend actually exposes those public bridge inputs
- if it reports 0 exported artifacts, bridge mode is enabled but has nothing useful to import yet
Quick check:
openclaw --version
openclaw wiki status
openclaw wiki doctor
openclaw wiki bridge import
If doctor says bridge has no public artifacts, that usually means the memory plugin side is not exposing them yet.
Docs:
If you paste the output of those 4 commands, I can tell you pretty quickly whether bridge mode is actually healthy or just enabled.
🦞 OpenClaw 2026.4.12 (1c0672b) — If something's on fire, I can't extinguish it—but I can write a beautiful postmortem.
Wiki vault mode: bridge
Vault: ready (/home/jimmy/.openclaw/wiki/main)
Render mode: native
Obsidian CLI: missing
Bridge: enabled (0 exported artifacts)
Unsafe local: disabled
Pages: 1 sources, 0 entities, 0 concepts, 1 syntheses, 6 reports
Source provenance: 0 native, 0 bridge, 0 bridge-events, 0 unsafe-local, 1 other
Warnings:
- Bridge mode is enabled but the active memory plugin is not exporting any public memory artifacts yet.
jimmy@jimmy-claw:~$ openclaw wiki doctor
🦞 OpenClaw 2026.4.12 (1c0672b) — I speak fluent bash, mild sarcasm, and aggressive tab-completion energy.
Wiki doctor: 1 issue(s) found
Wiki vault mode: bridge
Vault: ready (/home/jimmy/.openclaw/wiki/main)
Render mode: native
Obsidian CLI: missing
Bridge: enabled (0 exported artifacts)
Unsafe local: disabled
Pages: 1 sources, 0 entities, 0 concepts, 1 syntheses, 6 reports
Source provenance: 0 native, 0 bridge, 0 bridge-events, 0 unsafe-local, 1 other
Warnings:
- Bridge mode is enabled but the active memory plugin is not exporting any public memory artifacts yet.
Suggested fixes:
- Use a memory plugin that exports public artifacts, create/import memory artifacts first, or switch the wiki back to isolated mode.
jimmy@jimmy-claw:~$ openclaw wiki bridge import
🦞 OpenClaw 2026.4.12 (1c0672b) — I'll do the boring stuff while you dramatically stare at the logs like it's cinema.
Bridge import synced 0 artifacts across 0 workspaces (0 new, 0 updated, 0 unchanged, 0 removed). Indexes not refreshed (no-import-changes).
jimmy@jimmy-claw:~$
Yep, this output is pretty clear:
bridge mode itself is working
but there is nothing for it to import.
The key lines are:
Bridge: enabled (0 exported artifacts)Bridge import synced 0 artifactsactive memory plugin is not exporting any public memory artifacts yet
So the blocker is not the wiki vault.
It’s the memory side.
A few useful notes:
- Obsidian CLI missing is irrelevant here
- your wiki vault is healthy
- bridge mode is enabled correctly
- it just has no upstream artifacts
From current source, bridge depends on the active memory plugin exposing public artifacts. memory-core definitely does this. If you’re using another memory backend, it may simply not export bridge inputs yet.
If you are using memory-core, the bridge-importable artifacts are things like:
MEMORY.md/memory.md- files under
memory/ - memory event logs
So practical next step:
- check which memory backend/plugin you’re using
- if it’s
memory-core, make sure you actually have memory artifacts/files to export - rerun
openclaw wiki bridge import - if not, use
isolatedmode instead
If you paste your memory config/backend, I can tell you right away whether bridge should work with that setup or not.
command?
Yep, for the memory backend:
openclaw config get memory.backend
And for a quick memory health/status check:
openclaw memory status
Paste both outputs here and I’ll tell you whether bridge mode should work with your setup.
🦞 OpenClaw 2026.4.12 (1c0672b) — Gateway online—please keep hands, feet, and appendages inside the shell at all times.
Config path not found: memory.backend
That’s actually useful, not bad news.
Config path not found: memory.backend usually just means you haven’t set that key explicitly.
Two better checks:
openclaw config get plugins.slots.memory
openclaw memory status
But even without that, your earlier output already tells the main story:
- bridge mode is enabled
- the wiki is healthy
- there are no memory artifacts to import yet
So if you’re on the default memory setup, bridge will stay empty until there’s something like:
MEMORY.md/memory.md- files under
memory/ - memory event output
Then run:
openclaw wiki bridge import
If you paste openclaw memory status, I can narrow it down the rest of the way.
🦞 OpenClaw 2026.4.12 (1c0672b) — One CLI to rule them all, and one more restart because you changed the port.
Config path not found: plugins.slots.memory
🦞 OpenClaw 2026.4.12 (1c0672b) — I'll refactor your busywork like it owes me money.
Memory search disabled.
Memory search disabled.
Memory search disabled.
Memory search disabled.
Memory search disabled.
Memory search disabled.
Ah, got it, this is the real signal:
plugins.slots.memorymissing is usually fine, that just means it isn’t explicitly set- the real problem is
Memory search disabled.
So right now your setup is basically:
- wiki bridge = enabled
- memory = not active enough/usefully configured
- result = 0 artifacts to import
I’d do this next:
openclaw memory status --deep --agent main
That should give a more specific reason.
If it says no provider / embeddings aren’t configured, the usual fix is:
- configure memory/provider credentials
- run
openclaw memory index --force - run
openclaw wiki bridge import
If you want the quick path, try:
openclaw configure --section model
Then rerun the two commands above.
If bridge is your goal, you may also find isolated simpler unless you specifically want wiki pages built from active memory.