memcp

memcp is a Python MCP server that provides persistent external memory for Claude Code. It stores insights in SQLite (including a MAGMA 4-graph with semantic/temporal/causal/entity edges) and larger context content on disk, exposing 24 MCP tools for remembering/recalling, context-as-variable management, searching, graph traversal, cognitive reinforcement/consolidation, retention lifecycle operations, and project/session organization. It also supports pre-compact auto-save hooks and can deploy RLM-style sub-agents for analysis/mapping/synthesis/entity extraction.

Evaluated Mar 30, 2026 (21d ago)
Homepage ↗ Repo ↗ Infrastructure mcp claude-code memory persistent-storage sqlite knowledge-graph search context-management python
⚙ Agent Friendliness
64
/ 100
Can an agent use this?
🔒 Security
47
/ 100
Is it safe for agents?
⚡ Reliability
31
/ 100
Does it work consistently?

Score Breakdown

⚙ Agent Friendliness

MCP Quality
85
Documentation
70
Error Messages
0
Auth Simplicity
95
Rate Limits
0

🔒 Security

TLS Enforcement
60
Auth Strength
35
Scope Granularity
20
Dep. Hygiene
55
Secret Handling
70

README claims secret detection blocks accidental storage of API keys/tokens/credentials using regex patterns. However, no explicit auth model, transport security details, or fine-grained authorization are documented in the provided content; if exposed beyond localhost, additional safeguards would be needed.

⚡ Reliability

Uptime/SLA
0
Version Stability
45
Breaking Changes
40
Error Recovery
40
AF Security Reliability

Best When

You want Claude Code to retain and retrieve knowledge across long-lived workflows, and you can run a local MCP server with on-disk storage (SQLite + filesystem).

Avoid When

You cannot accept local persistence of prompts/insights, or you require centralized access controls and auditing beyond local deployment.

Use Cases

  • Persisting decisions, findings, and preferences across Claude Code sessions
  • Storing large documents externally and loading only needed portions via chunk/peek/filter tools
  • Linking related insights through semantic/temporal/causal/entity graph relationships
  • Running intent-aware recall (e.g., why/when questions) using graph edges
  • Implementing retention/archival workflows for memory and contexts
  • Multi-project memory separation and session tracking

Not For

  • Applications needing a hosted/managed cloud service with SLAs and guaranteed uptime
  • Use cases requiring strong, enterprise-grade auth/tenant isolation provided by a remote API (this is locally operated)
  • Scenarios where storing potentially sensitive data in local SQLite/filesystem is unacceptable
  • Environments that disallow local background indexing/embedding computation

Interface

REST API
No
GraphQL
No
gRPC
No
MCP Server
Yes
SDK
No
Webhooks
No

Authentication

Methods: Local MCP server registration via Claude Code CLI (no documented network auth in the provided content)
OAuth: No Scopes: No

The README focuses on local installation and MCP registration; no explicit auth mechanism (API keys/OAuth) is described for tool calls. Security relies primarily on local execution and filesystem access controls.

Pricing

Free tier: No
Requires CC: No

Open-source (MIT). Costs are indirect: local compute/storage and optional embedding/vector dependencies.

Agent Metadata

Pagination
none
Idempotent
False
Retry Guidance
Not documented

Known Gotchas

  • Requires correct MCP server registration and, optionally, deployment of sub-agents/hooks to the Claude Code environment; misconfiguration can cause missing functionality
  • Some capabilities depend on optional extras (semantic/fuzzy/vector/hnsw/ner/async/llm); without them, behavior may degrade

Alternatives

Full Evaluation Report

Comprehensive deep-dive: security analysis, reliability audit, agent experience review, cost modeling, competitive positioning, and improvement roadmap for memcp.

AI-powered analysis · PDF + markdown · Delivered within 30 minutes

$99

Package Brief

Quick verdict, integration guide, cost projections, gotchas with workarounds, and alternatives comparison.

Delivered within 10 minutes

$3

Score Monitoring

Get alerted when this package's AF, security, or reliability scores change significantly. Stay ahead of regressions.

Continuous monitoring

$3/mo

Scores are editorial opinions as of 2026-03-30.

8642
Packages Evaluated
17761
Need Evaluation
586
Need Re-evaluation
Community Powered