Agent Memory MCP server
Commit Nov 14, 2026Gives coding agents persistent local memory and semantic search over decisions, facts, and project docs stored in SQLite.
Also known as ipiton/agent-memory-mcp, agent-memory-mcp
- Stars
- 45
- Tools
- —
- Downloads, 30 days
- —
- Latest version
- 0.18.0
- Last commit
- Nov 14, 2026
- License
- MIT
What is Agent Memory MCP?
Agent Memory gives coding agents persistent local memory and semantic search over decisions, facts, and project docs stored in SQLite.
Summary generated from the project's README
About Agent Memory MCP server
Providing coding agents with persistent local memory and semantic search over project documentation is what Agent Memory MCP achieves. It allows developers to store decisions, facts, and procedures in a local SQLite database, making them searchable alongside ADRs and runbooks. This setup ensures that an agent can recall the reasoning behind specific architectural choices or find relevant runbooks during an incident.
Users can install this as a Claude Code plugin for a zero-setup experience or as a standalone server for tools like Cursor, Windsurf, or VS Code. While it operates locally with one binary and a database per project, an HTTP mode is available for shared team instances. Standalone installations require an embedding provider to function. The system includes specialized tools like steward_run to manage memory quality and drift_scan to identify when stored information no longer matches the current state of the repository.
What you can do with it
- 1Recalling architectural decisionsAsk an AI agent to explain the reasoning behind a specific configuration or scaling decision.
- 2Searching project documentationFind specific runbooks or procedures by performing a semantic search across all project docs.
- 3Maintaining memory accuracyIdentify when stored facts have become outdated due to changes in the codebase.
- 4Cleaning up duplicate informationUse automated stewardship to find and resolve contradictions or stale entries in the memory.
Written from the project's README. Every use case is checked against a quote from it.
Automated checks
Ran Nov 15, 2026Full report →Not verified yet. A server needs a score of 80, no serious finding, and enough data for us to check. How checks work · What to fix for more points
- Source repository linked and reachableipiton/agent-memory-mcp is public and reachablegithub.com/ipiton/agent-memory-mcp ↗Pass
- Package names this repository as its sourceNot checked
- Build provenance (Sigstore attestation)Not checked
- Open source license (OSI)MITspdx.org/licenses/MIT.html ↗Pass
- Commit or release in the last yearLast commit or release 2026-11-14github.com/ipiton/agent-memory-mcp ↗Pass
- OpenSSF ScorecardNot checked
- Package name not close to a popular packageNot checked
- No critical advisory in the package or its dependenciesNot checked
- No hidden instructions to the model0 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
- Destructive tools are labelledNot checked
- Remote endpoint answers the MCP handshakeNot checked
README
View on GitHubagent-memory-mcp
Long-term memory and project-docs search for coding agents, in one local MCP server backed by SQLite.
Your agent remembers why a decision was made, finds the runbook that matches an incident, and notices when what it remembers no longer matches the repo.
Install
Claude Code. Nothing else to set up: no service, no model, one local database per project. Details.
/plugin install agent-memory --marketplace ipiton/agent-memory-mcp
Codex, Cursor, Gemini CLI, OpenCode, VS Code, Windsurf, or Claude Code without the plugin. A standalone server, which needs one embedding provider. Details, including builds without Homebrew.
brew tap ipiton/tap
brew install agent-memory-mcp
cd your-repo
agent-memory-mcp setup codex --yes
Restart the agent afterwards.
What it does
- Searches memory and docs together. Decisions, facts and procedures are ranked alongside your docs, ADRs and runbooks, and every hit says how far to trust it.
- Keeps memory clean.
steward_runfinds duplicates, contradictions and stale entries, and sends anything risky to a review inbox. - Notices drift.
drift_scanflags memories whose source file changed or disappeared. - Leaves the last word to you. Notes become canonical only through
promote_to_canonical. An agent cannot verify its own notes. - Stays local. One binary and SQLite. An HTTP mode is there for a shared team instance.
> How many replicas should payments-api run, and why?
1. **Why payments-api runs 4 replicas** (relevance: 0.79)
Trust: layer=raw source=decision confidence=0.96 freshness=1.00 owner=engineering
Content: Decision 2026-09-12: payments-api runs 4 replicas, not 2. Two replicas dropped
requests during the September card-network spike. Source: ./deploy/values.yaml.
More in Features.
Documentation
| Topic | Where |
|---|---|
| Claude Code plugin | PLUGIN.md |
| Standalone install, connecting agents, team service | INSTALLATION.md |
| Tools and CLI | MCP_TOOLS.md, CLI.md |
| Configuration | CONFIGURATION.md |
| Stewardship and memory lifecycle | STEWARDSHIP.md |
| Architecture and security | ARCHITECTURE.md, SECURITY.md |
| Troubleshooting and changes | TROUBLESHOOTING.md, CHANGELOG.md |
Development
make build
make test
make vet
Conventions are in CONTRIBUTING.md. License: MIT.
Agent Memory MCP questions
Does agent-memory-mcp require a cloud service or model?
No. The Claude Code plugin needs nothing else to set up: no service, no model, and one local database per project. The standalone server only needs one embedding provider.
Can the agent write memories straight into canonical memory?
No. Notes become canonical only through the promote_to_canonical tool, since an agent cannot verify its own notes.
How does agent-memory-mcp detect outdated memories?
Its drift_scan tool flags memories whose source file changed or disappeared.
Can memory and project docs be searched together?
Yes. Decisions, facts and procedures are ranked alongside docs, ADRs and runbooks, and each hit reports how far to trust it.
How do I install agent-memory-mcp for Codex or Cursor?
Use the standalone server: tap and install with Homebrew, then run agent-memory-mcp setup with your agent name and restart the agent.