agentmem

Install

Pick your client. Read the repository before you give any tool access to your data.

About

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this server can do and what it asks for, from its tool list and install config. Checked Oct 25, 2026.

Tools

9 read · 6 write · 1 delete

Secrets

None asked for

Install scripts

Not scanned yet

Tools (16)

Read from the source code; the running server may differ. Descriptions are the server’s own words; hidden characters are shown as [U+…].

• add_procedure

  Add a behavioral rule (procedural memory). Procedural memories are never auto-expired or auto-compacted. They are meant to be injected into the agent's system prompt to modify behavior. Examples: 'Always respond in bullet points', 'Never expose API keys'.

• compact

  Archive low-value memories to reduce noise in search results. Archives memories older than max_age_days with access_count <= min_access. Core and procedural tier memories are never auto-archived. Use dry_run=true to preview how many would be archived.

• consolidate

  Find and merge semantically similar memories (near-duplicates). Groups memories with cosine similarity above the threshold, keeps the longest/newest in each group, and archives the rest. Requires an embedding function to be set. Use dry_run=true to preview groups without modifying anything.

• entities

  List all known entities extracted from memories. Shows entity names, types, and how many memories reference each. Useful for discovering what entities are stored in memory.

• forget

  Archive a memory (soft delete). The memory is hidden from search but not permanently deleted.

• get_procedures

  Get all active procedural memories (behavioral rules) formatted for system prompt injection. Returns a markdown-formatted string with all rules as a bullet list. If no procedural memories exist, returns empty string. Use this at the start of a session to load agent behavioral rules.

• history

  Get the version history of a memory. Follows the supersedes chain to show all previous versions. Returns newest first.

• process_conversation

  Automatically extract and store memories from a conversation history. Scans messages for facts, decisions, preferences, TODOs, config values, learnings, and important notes using regex heuristics (no LLM). Stores extracted memories with appropriate tiers and tags. Accepts standard OpenAI/Anthropic message format.

• recall

  Search agent memory using hybrid keyword + semantic search. Use this to find information from past sessions, stored facts, learned knowledge, or working state. Returns the most relevant memories ranked by relevance.

• related

  Find memories connected to the same entity. Entities are auto-extracted from memory content: @mentions, URLs, IPs, ports, file paths, env vars, money, etc. Use this to discover all memories referencing a specific entity.

• remember

  Store a new memory. Use this to save important facts, learnings, decisions, or any information that should persist across sessions. Automatically deduplicates by content.

• save_state

  Emergency save of current working context. Call this before context compression to preserve your current state. Replaces any previous working state.

• stats

  Get memory statistics: total count, breakdown by tier, database size, and latest activity.

• today

  Get all memories created today. Use this to see what happened during the current day.

• unarchive

  Restore an archived memory, making it active and searchable again.

Signals

Stars

6

Forks

1

Open issues

0

Last commit

Feb 28, 2026 (8 months ago)

Last release

Feb 21, 2026 (8 months ago)

From GitHub (fetched Oct 24, 2026).

Details

Registry name

io.github.oxgeneral/agentmem

Version

0.3.0

License

MIT

Repository

github.com/oxgeneral/agentmem

First seen

Oct 24, 2026

Last checked

Oct 24, 2026

Source

Official MCP Registrygithub-graphql