Claude-Mem Search (Cowork)
SkillSearchThis skill lets an AI agent search the user's persistent Claude-Mem memory on cmem.ai, which stores timestamped observations from past sessions across their agents. When someone asks what the agent remembers about a project, decision, file, or task, the skill runs a bundled CLI to find those earlier observations and answer in plain language. It also includes a status command for diagnosing missing API keys or endpoint errors.
Then ask your AI: use the Claude-Mem Search (Cowork) skill
Details
Instructions available. Your AI can read the instructions. Execution depends on the setup they require.
Account requirements not reviewed. Check the skill instructions before use; Ahel provides instructions and does not run this skill.
No other account needed.
Have a Claude-Mem account on cmem.ai and an API key for it.
What your AI can do with it
- Search Claude-Mem memory with the bundled cmem-hook.mjs CLI
- Run broad keyword searches across past session observations
- Narrow results with specific terms and a smaller limit
- Synthesize a plain-language answer with timestamps when asked when
- Diagnose missing API keys or endpoint errors with the status command
Getting started
- Have a Claude-Mem account on cmem.ai and an API key for it.
- Add the cmem.ai API key to the plugin's configuration.
- Run the bundled CLI with a search query and a limit to look for past observations.
- If searches return nothing or error, run the status command and read its output.
What this skill tells your AI
The instructions your AI receives, as published by thedotmack/claude-mem in cowork/skills/mem-search/SKILL.md and read by Ahel’s review.
Search the user's persistent Claude-Mem memory on cmem.ai. Memory contains timestamped observations synthesized from past sessions across all their agents (Claude Code, Cowork, Codex, and others).
How to search
Run the bundled CLI (no dependencies, uses the plugin's configured API key):
node "${CLAUDE_PLUGIN_ROOT}/scripts/cmem-hook.mjs" search "your query" --limit 20
Progressive search method
Follow claude-mem's Index → Timeline → Transcript discipline — cheap passes first, expensive detail only for confirmed hits:
- Index pass — run 1–3 broad keyword searches (project names, file names, error strings, feature names). Skim titles/summaries only.
- Narrow pass — re-search with the most specific terms found in step 1
(IDs, exact phrases) and a smaller
--limit. - Answer — synthesize from the observations returned. Quote timestamps when the user asks "when".
Do not dump raw search output at the user; extract the relevant observations and answer in plain language.
Diagnostics
If searches return nothing or error:
node "${CLAUDE_PLUGIN_ROOT}/scripts/cmem-hook.mjs" status
Report the status output plainly: a MISSING api key means the plugin needs the
user's cmem.ai key added to its configuration; a 404 on /api/hooks/context
just means the newer Pro endpoint isn't deployed — search still works via
/api/mcp.
Notes
- Some filters (date ranges, type) may be silently ignored by the cloud API until MCP parity ships — prefer keyword narrowing over filter flags.
- Never write secrets into search queries; queries are sent to cmem.ai.
Signals
- GitHub stars
- 99k
- Forks
- 9k
- Last commit
- Oct 2026
Questions
- What do you remember about X?
- The skill searches Claude-Mem memory for observations about X and answers in plain language. It runs broad keyword searches first, then narrower ones, and quotes timestamps when you ask when something happened.
- What did we do last session?
- It searches saved observations from past sessions and summarizes the relevant work. Results depend on what was stored in Claude-Mem memory, and it can quote timestamps for when things happened.
- How do I search memory?
- Ask the agent to search memory, or run the bundled CLI with a query and a limit. The skill follows an index pass, a narrow pass, and then a synthesized answer.
Advanced
- Item type
- skill
- Key
mem-search-thedotmack- Source
- github.com/thedotmack/claude-mem