#How do stop openclaw from forgetting

1 messages · Page 1 of 1 (latest)

stray crypt
#

how do i stop openclaw from forgetting details and projects

turbid raftBOT
#

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> âś… Mark Solution

swift scarabBOT
#

OpenClaw only “remembers” what gets written to its workspace files — there isn’t hidden magic memory.

Do this:

  1. Put durable facts in the workspace

    • ~/.openclaw/workspace/MEMORY.md → stable long-term facts/preferences/decisions
    • ~/.openclaw/workspace/memory/YYYY-MM-DD.md → daily/project running notes
    • AGENTS.md / USER.md / TOOLS.md → standing instructions and context

  2. Tell it explicitly
    Say things like:

    “Remember this project: … write a project brief to memory.”
    “Before we stop, summarize the current state into memory.”

  3. Check memory indexing

# 
   openclaw memory status
   openclaw memory index --force
   openclaw memory search "your project name"

  1. Use the same agent/workspace/session
    If you switch agents, channels, or workspaces, it may look like it forgot. Check:

    openclaw config get agents.defaults.workspace

    and in chat use /status or /context list.

  2. For more automatic recall, enable Active Memory
    This makes OpenClaw search memory before replying. Docs: https://docs.openclaw.ai/concepts/active-memory

Main docs:

stray crypt
#

OpenClaw 2026.4.26 (be8c246) / Memory Search (main)
Provider: none (requested: auto)
Model: none
Sources: memory
Indexed: 13/13 files · 55 chunks
Dirty: yes
Store: ~/.openclaw/memory/main.sqlite
Workspace: ~/.openclaw/workspace
Dreaming: off
Embeddings: unavailable
Embeddings error: No GitHub token available for Copilot embedding provider

No API key found for provider "openai". Auth store: /Users/warren/.openclaw/agents/main/agent/auth-profiles.json (agentDir: /Users/warren/.openclaw/agents/main/agent). Configure auth for this agent (openclaw agents add <id>) or copy auth-profiles.json from the main agentDir.

No API key found for provider "google". Auth store: /Users/warren/.openclaw/agents/main/agent/auth-profiles.json (agentDir: /Users/warren/.openclaw/agents/main/agent). Configure auth for this agent (openclaw agents add <id>) or copy auth-profiles.json from the main agentDir.

No API key found for provider "voyage". Auth store: /Users/warren/.openclaw/agents/main/agent/auth-profiles.json (agentDir: /Users/warren/.openclaw/agents/main/agent). Configure auth for this agent (openclaw agents add <id>) or copy auth-profiles.json from the main agentDir.

No API key found for provider "mistral". Auth store: /Users/warren/.openclaw/agents/main/agent/auth-profiles.json (agentDir: /Users/warren/.openclaw/agents/main/agent). Configure auth for this agent (openclaw agents add <id>) or copy auth-profiles.json from the main agentDir.

No API key found for provider "bedrock". AWS credentials are not available. Set AWS_ACCESS_KEY_ID/AWS_SECRET_ACCESS_KEY, AWS_PROFILE, or AWS_BEARER_TOKEN_BEDROCK, configure an EC2/ECS/EKS role, or set agents.defaults.memorySearch.provider to another provider.
By source:
memory · 13/13 files · 55 chunks
Vector: unknown
FTS: ready
Embedding cache: enabled (0 entries)
Batch: disabled (failures 0/2)
Recall store: 9 entrie

swift scarabBOT
#

Yep — this output explains the “forgetting” feeling.

Your memory exists:

Indexed: 13/13 files · 55 chunks
FTS: ready

…but semantic memory is not working:

Provider: none
Embeddings: unavailable

So OpenClaw can only do keyword-ish search, not “same meaning, different wording” recall.

#

Do this:

openclaw memory index --force

Then configure one embedding provider for the main agent. If you have GitHub Copilot:

openclaw models auth --agent main login-github-copilot

Or use OpenAI/Gemini/etc:

openclaw models auth --agent main add