anima-recall-mcp 0.1.2

Local-first decision memory and consistency review MCP server.

ANIMA Recall MCP

ANIMA Recall is a local-first MCP server for preserving decisions and reviewing new choices against relevant past evidence. Memory content and comparison logic stay on the user's machine.

Tools

Tool	Purpose	Effect
anima_remember	Store an explicit decision or principle	Writes
anima_recall_memories	Find relevant memories	Reads
anima_compare_choice	Surface possible conflicts with past decisions	Reads
anima_recent_memories	List recent memories	Reads
anima_reflect	Review clusters, conflicts, and promotion candidates	Preview by default
anima_forget	Delete one explicitly confirmed memory	Deletes

anima_reflect does not promote principles unless the caller explicitly sets promote=true. Its output is review support, not an authoritative judgment.

Install and run

python -m pip install anima-recall-mcp
$env:ANIMA_RECALL_DB="$PWD\data\anima_recall.db"
$env:ANIMA_RECALL_OWNER_PEPPER="replace-with-at-least-32-random-bytes"
anima-recall-mcp

• MCP endpoint: http://127.0.0.1:8000/mcp
• Health: http://127.0.0.1:8000/health
• Privacy notice: http://127.0.0.1:8000/privacy

Set the bundle license once for both ANIMA and handoffpack:

$env:ANIMA_LICENSE="ANIMA1...."

ANIMA_RECALL_LICENSE is also accepted as an ANIMA-specific override.

Privacy and deployment

• No user name or email is required in tool input.
• Memory content is stored in the configured local SQLite database.
• A high-entropy ANIMA_RECALL_OWNER_PEPPER is required for deployment.
• Multi-tenant gateways must set ANIMA_RECALL_GATEWAY_SECRET and authenticate requests with the documented gateway header.
• Back up the database before migrations or upgrades.

See docs/CONNECT.md, docs/KC_DEPLOYMENT.md, and docs/PLAYMCP_SUBMISSION.md for connection and deployment details.

Development

$env:PYTHONPATH="src"
python -m pytest -q
python -m build