About io.github.donggyun112/keymem
io.github.donggyun112/keymem is an MCP server in the Communication category: associative key-graph memory for LLM agents — recall by association, not vector similarity.
Install
claude mcp add io-github-donggyun112-keymem -- npx -y keymemnpx -y keymemThis server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.
Tools (16)
browse_keys(namespace, hubs_only?, limit?, offset?)Browse a namespace's active key vocabulary, hubs first, when recall has no entry hit.
Parameters not published.
cleanup_expired()Delete memories whose TTL has expired
Parameters not published.
confirm_memory(memory_id, evidence, namespace?, source?)Refresh freshness and deepen a current memory after explicit user evidence, an authoritative source, or direct observation. A read alone is not evidence.
Parameters not published.
correct(memory_id, content, keys?, key_types?, ttl_seconds?, decay_profile?, related_to?)Versioned update. The immediate predecessor is preserved but inactive; omitted TTL/profile inherit from it.
Parameters not published.
dismiss(memory_id, key_id, namespace?)Negative feedback: the fact is fine, this key should not have surfaced it. Weakens that one edge (floored, never severed) and cancels its pending alias learning
Parameters not published.
forget(memory_id)Permanently delete
Parameters not published.
get_conversation(session_id, turn?, agent?)Load original conversation turns from the host agent's on-disk transcript (Claude Code / Codex), normalized to `{turn, role, content, ts}`
Parameters not published.
list_memories(namespace?)List active memories with keys, depth, access count, and validity
Parameters not published.
list_sessions(agent?, limit?)Discover recent host-agent conversation sessions (Claude Code, Codex) on this machine, newest first
Parameters not published.
memory_stats()Get current key/memory/link counts
Parameters not published.
Show all 16 tools
read_key(key_id, query?, namespace?, limit?, offset?)List ranked memory IDs and metadata connected to one key. Pass the original query for relevance ordering; supports pagination for hubs.
Parameters not published.
read_memory(memory_id, via_key_id?, namespace?)Read full memory content, connected keys, and `validity`. Updates access and, when `via_key_id` is supplied, that traversed edge; never confirms or deepens content.
Parameters not published.
recall(query, top_k?, namespace?, context?, explain?)Return ranked keys plus one passive Top-1 memory with `validity`, `matched_key`, and `connected_keys`. `explain:true` also returns namespace counts.
Parameters not published.
related(memory_id)Find memories sharing keys (associative exploration)
Parameters not published.
remember(content, keys, key_types?, namespace?, ttl_seconds?, decay_profile?, related_to?)Save memory with key concepts, optional TTL, and a `transient`, `standard`, `stable`, or `permanent` decay profile.
Parameters not published.
remember_batch(items)Save multiple memories; each item accepts `ttl_seconds` and `decay_profile`
Parameters not published.
Ask AI
Ask AI about io.github.donggyun112/keymem
Powered by Claude · Grounded in docs
Security checks
6 checks have not been run yet.