MCP Integration
The mempill-mcp package is a FastMCP server that
exposes the mempill engine over the MCP stdio transport. It lets any MCP-compatible agent
(Claude Desktop, Claude Code, custom) read and write mempill claims without embedding the
Python wheel directly.
What mempill-mcp exposes
Section titled “What mempill-mcp exposes”The server registers four tools, implemented in
mempill-mcp/src/mempill_mcp/tools.py:
| Tool | Signature | Description |
|---|---|---|
ingest_claim |
(subject, predicate, value, provenance, ...) |
Write a belief claim |
query_memory |
(subject, predicate, as_of_tx_time?) |
Read the canonical belief |
reconcile |
(subject_lines) |
Trigger conflict reconciliation |
audit |
(limit?, claim_ref?, from_tx_time?) |
Read the immutable audit ledger |
These map directly to engine.ingest_claim, engine.query_memory, engine.reconcile,
and engine.query_audit. The agent_id is injected from the environment — the MCP
client never needs to pass it.
ingest_claim parameters
Section titled “ingest_claim parameters”subject str Entity (e.g. "user:alice")predicate str Property (e.g. "city")value any JSON-serialisable valueprovenance str | dict "External:UserAsserted" | "External:ExternalFirstHand" | "RecallReEntry" | "ModelDerived" | raw wire dictcardinality str = "Functional" "Functional" | "SetValued" | "Unknown"confidence_value float = 0.9 Value confidence [0, 1]confidence_valid_time float = 0.9 Temporal confidence [0, 1]criticality str = "Low" "Low" | "Medium" | "High" | "Critical"valid_time dict | None {"start"?: ISO-8601, "end"?: ISO-8601}derived_from list[str] | None Source claim UUIDsWhen disposition is non-committed (Contested, QueuedForAdjudication, etc.), the response
includes a human-readable status_reason field explaining why.
query_memory parameters
Section titled “query_memory parameters”subject strpredicate stras_of_tx_time str | None ISO-8601 UTC for point-in-time queryReturns {"belief": {...}}. Contested beliefs also include status_reason.
reconcile parameters
Section titled “reconcile parameters”subject_lines list[list[str]] Each inner list is [subject, predicate]Example: [["user:alice", "city"], ["user:bob", "location"]]
audit parameters
Section titled “audit parameters”limit int = 50claim_ref str | None UUID of a specific claim to filterfrom_tx_time str | None ISO-8601 UTC lower boundEnvironment variables
Section titled “Environment variables”| Variable | Required | Default | Meaning |
|---|---|---|---|
MEMPILL_AGENT_ID |
Yes | — | Agent identity for all writes in this session. Must be set before launch — the server fails fast with a clear error if missing. |
MEMPILL_DB_PATH |
No | in-memory | Absolute path to the SQLite database file. If omitted the server uses an ephemeral in-memory engine (data lost on exit). |
MEMPILL_AGENT_ID is the single-writer key: all claims written in a session belong to this
agent. Use a stable, meaningful identifier (e.g. "claude-desktop-alice").
Installation
Section titled “Installation”The mempill wheel is on PyPI; mempill-mcp is installed from source. Set up both into a venv:
# 1. Install the published mempill wheelpip install mempill
# 2. Install the MCP adapter from sourcecd mempill/mempill-mcppip install .
# 3. Test the serverMEMPILL_AGENT_ID=test python -m mempill_mcp# Should start and wait for MCP stdio input; Ctrl-C to exit.The easiest approach is to use the mempill-demo repository, which ships a pre-built venv
with all dependencies. See Examples for the demo.
Claude Desktop configuration
Section titled “Claude Desktop configuration”Claude Desktop reads mcpServers from its config file. Copy the block below, replace the
absolute paths, and restart Claude Desktop (full quit — menu bar → Quit, not just close):
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
{ "mcpServers": { "mempill": { "command": "/ABSOLUTE/PATH/TO/mempill-demo/.venv/bin/python", "args": ["-m", "mempill_mcp"], "env": { "MEMPILL_AGENT_ID": "my-agent", "MEMPILL_DB_PATH": "/ABSOLUTE/PATH/TO/mempill-demo/demo.db" } } }}Important notes:
- Use the
pythonbinary from the project venv that hasmempillandmempill-mcpinstalled. Launchingmempill-mcpas a bare command requires both packages to be on the system Python, which is unlikely. The venv python is the reliable path. - Claude Desktop does not expand
$HOMEor shell variables in this file — use literal absolute paths. - The
MEMPILL_DB_PATHpath is shared across all Claude Desktop sessions for this agent. Use distinct paths per agent identity if you need per-agent isolation.
This config is sourced from mempill-demo/mcp/claude_desktop_config.json.example.
Claude Code (.mcp.json) configuration
Section titled “Claude Code (.mcp.json) configuration”Claude Code reads .mcp.json from the project root. This file is gitignored in mempill-demo.
Claude Code supports ${VAR:-default} expansion in the env block:
{ "mcpServers": { "mempill": { "type": "stdio", "command": "/ABSOLUTE/PATH/TO/mempill-demo/.venv/bin/python", "args": ["-m", "mempill_mcp"], "env": { "MEMPILL_AGENT_ID": "${MEMPILL_AGENT_ID:-demo-agent}", "MEMPILL_DB_PATH": "${MEMPILL_DB_PATH}" } } }}If MEMPILL_DB_PATH is unset in the shell environment, the server opens an in-memory
(ephemeral) engine.
This config is sourced from mempill-demo/mcp/.mcp.json.example.
How the server lifecycle works
Section titled “How the server lifecycle works”The server opens the mempill engine once in a FastMCP lifespan context manager and
yields {"engine": engine, "agent_id": agent_id} to all tool functions. This means:
- Engine startup cost (migrations, WAL setup) happens once per server process, not per call.
- All four tools share the same engine instance and agent identity.
- SQLite WAL mode is active — concurrent reads from other processes do not block.
- When
MEMPILL_DB_PATHis set, the database file persists after the server exits.
Testing the server without an MCP host
Section titled “Testing the server without an MCP host”mempill-demo/mcp/verify_stdio.py is a small script that sends a tool-list request over
stdio and validates the response. Run it against the installed server:
echo '{"jsonrpc":"2.0","id":1,"method":"tools/list"}' \ | MEMPILL_AGENT_ID=test .venv/bin/python -m mempill_mcpYou should see a JSON response listing all four tools.
Contested beliefs in MCP context
Section titled “Contested beliefs in MCP context”When query_memory returns a Contested belief, the response includes:
{ "belief": { "status": "Contested", "primary": null, "alternatives": [ {"fact": {"value": "Berlin"}}, {"fact": {"value": "Munich"}} ] }, "status_reason": "This claim conflicts with one or more existing beliefs..."}A well-behaved MCP agent should surface this to the user rather than silently picking one value. See Concepts: Contested and Dispositions.
Related guides
Section titled “Related guides”- Writing an Oracle — adding HITL resolution to MCP sessions
- Python (Advanced) — embedding the wheel directly instead of MCP
- Examples — the demo repo with a working MCP config