Imported from percy-raskova/kimi-claude-mem (skills/using-claude-mem/SKILL.md). Install upstream with npx skills add percy-raskova/kimi-claude-mem --skill using-claude-mem. Copyright stays with the author.

Using claude-mem memory

Name: using-claude-mem
Author: percy-raskova

This Kimi Code installation has persistent memory via claude-mem. Your past sessions are captured as observations (decisions, bugfixes, features, discoveries) and searchable through the mcp__claude-mem__* MCP tools.

3-layer workflow (always follow)

search(query) → index of compact results with IDs (~50-100 tokens each)
timeline(anchor=ID) → context around an interesting result
get_observations([IDs]) → full details ONLY for the IDs you filtered to

Never fetch full details without filtering first — it wastes 10x the tokens.

When to use

"What did we do last week / how did we solve X before?" → search memory first
Starting work in a familiar project → the context digest injected at prompt submit already covers recent highlights; search only for older or specific topics
Before recommending an approach → check whether a past session already tried it

Hygiene

Memory contains what past sessions observed; verify anything load-bearing against the actual code before acting on it.
Never write secrets, tokens, or credentials into prompts you want remembered — everything is stored locally in ~/.claude-mem/.

# after Install: the listing is in your workspace's routing pool - a plan picks it for its slot curl -s -X POST https://api.opensmartroute.ai/api/v1/route -H 'Authorization: Bearer $OSR_API_KEY' -H 'Content-Type: application/json' -d '{"text": "...", "plan": true}'

{ "ocm": "1", "id": "percy-raskova-kimi-claude-mem-using-claude-mem", "kind": "skill", "name": "using-claude-mem", "description": "How to use the claude-mem memory tools (mcp__claude-mem__*) to recall past sessions — search, timeline, then fetch details. Load at session start.", "publisher": "percy-raskova", "version": "1.0.0", "capabilities": { "domains": [ "general" ], "tags": [ "skill-md", "github" ], "languages": [ "en" ] }, "quality_prior": 0.6, "examples": [ "How to use the claude-mem memory tools (mcp__claude-mem__*) to recall past sessions — search, timeline, then fetch details. Load at session start." ], "primary": false, "metadata": { "source": { "provider": "github", "repository": "https://github.com/percy-raskova/kimi-claude-mem", "path": "skills/using-claude-mem/SKILL.md", "ref": "3d63b9e9b910901cec78f83a41ef904731388b27", "url": "https://github.com/percy-raskova/kimi-claude-mem/blob/3d63b9e9b910901cec78f83a41ef904731388b27/skills/using-claude-mem/SKILL.md", "key": "percy-raskova/kimi-claude-mem/skills/using-claude-mem/SKILL.md" } }, "instructions": "# Using claude-mem memory\n\nThis Kimi Code installation has persistent memory via claude-mem. Your past\nsessions are captured as observations (decisions, bugfixes, features,\ndiscoveries) and searchable through the `mcp__claude-mem__*` MCP tools.\n\n## 3-layer workflow (always follow)\n\n1. `search(query)` → index of compact results with IDs (~50-100 tokens each)\n2. `timeline(anchor=ID)` → context around an interesting result\n3. `get_observations([IDs])` → full details ONLY for the IDs you filtered to\n\nNever fetch full details without filtering first — it wastes 10x the tokens.\n\n## When to use\n\n- \"W", "cost": { "context_tokens": 297 } }

using-claude-mem

About

Using claude-mem memory

3-layer workflow (always follow)

When to use

Hygiene

Use it

Manifest

Reviews