← ClaudeAtlas

tdai-memorylisted

Long-term memory for coding agents. Automatically recall project context before answering, and capture decisions, learnings, and fixes after completing work. Use when the user references past work, starts a new session, or when the task needs project context that is not in the current conversation.
tinhien11/tdai-memory-mcp · ★ 4 · AI & Automation · score 78
Install: claude install-skill tinhien11/tdai-memory-mcp
You have access to a long-term memory server via MCP. Use the tools automatically as described below. Do not ask the user for permission to use memory. ### Core tools (always available) `recall` `capture` `search` `forget` `resolve` `handoff` `adr` `update` `consolidate` ### Advanced tools (only when `TDAI_ENABLE_ADVANCED=1`) **CodeGraph:** `codegraph_index` `codegraph_search` `codegraph_callers` `codegraph_callees` `codegraph_impact` `codegraph_list` **Wiki:** `wiki_ingest` `wiki_search` `wiki_get` `wiki_outdated` **Knowledge:** `knowledge_create` `knowledge_get` `knowledge_list` `knowledge_delete` **Skill:** `skill_get` `skill_list` `skill_search` ## When to recall Call `recall` at the start of a session, or when the user references past work. Do this BEFORE you answer or start coding. ``` recall({ "query": "<the user's question or task summary>", "mode": "hybrid" }) ``` If the user says any of these, call `recall` first: - "do you remember" - "last time we" - "what did we decide about" - "have we seen this error before" - "continue from where we left off" - any reference to a previous session or past decision If recall returns results, use them to inform your answer. If recall returns "No memory found", proceed normally. Do not mention that recall returned nothing. ## Multi-tenant isolation If the user works in a team context, pass `team_id`, `agent_id`, `user_id`, or `task_id` to isolate memory. When you set `team_id`, all queries filter by that value. This