Agent Memory Mcp
Gives Claude a persistent memory layer with provenance tracking and time-weighted recall. Stores memories via memory_store with confidence scores and source metadata, retrieves them with memory_recall using decay-weighted ranking (30-day half-life), and runs a feedback loop through memory_feedback to surface what actually helps over time. Backed by local SQLite, so no API keys or remote dependencies. Reach for this if you're building an agent that runs across sessions and you're tired of watching it rediscover the same facts or waste context on stale information. The developer reports dropping session startup from 31K tokens to 800 in production use. Categories include lessons, strategies, preferences, and operational knowledge.
- tools
- 4
- transport
- STDIO
- env var
- 1
- last updated
- Mar 20, 2026
Tools
4 toolsPublic tool metadata for what this MCP can expose to an agent.
memory_storeStore a memory with provenance metadata.
Store a memory with provenance metadata.
No parameter schema in public metadata yet.
memory_recallRetrieve memories by keyword query and/or category, ranked by decay-weighted relevance.
Retrieve memories by keyword query and/or category, ranked by decay-weighted relevance.
No parameter schema in public metadata yet.
memory_feedbackRecord whether a recalled memory was useful. This is the flywheel.
Record whether a recalled memory was useful. This is the flywheel.
No parameter schema in public metadata yet.
memory_statsGet counts and averages for the memory store.
Get counts and averages for the memory store.
No parameter schema in public metadata yet.
Sponsors
Agent CaffeineSponsoredClose your lid, come back to finished work
SnugSponsoredThe clipboard manager for the AI eraPromote your product hereReach 125k+ visitors monthly
WaxSponsoredBlock anything with supervision, without erasing your iPhonePromote your product hereReach 125k+ visitors monthly
knobsSponsoredSee your site on any phone, any day, in any languageConfiguration
MEMORY_DB_PATHPath to SQLite database file (default: ~/.agent-memory/memory.db)
Readme
agent-memory-mcp
MCP server for agent memory with provenance tracking, decay-weighted recall, and feedback loops.
Most agent memory systems treat memories as free-floating facts. This one tracks where each memory came from, how confident you should be in it, and whether it was actually useful — so your agent stops rediscovering the same things and starts getting smarter over time.
Why this exists
Agents waste tokens. A lot of them. Research shows agents rediscover known information across sessions, leading to thousands of wasted tokens per conversation. Flat files are auditable but unsearchable. Vector DBs have great recall but no staleness signals. Structured state is brittle.
This is a memory layer that fixes the actual problems:
Details
- Repository
- github.com/kira-autonoma/agent-memory-mcp
- Category
- AI & LLM Tools
- Package
@kiraautonoma/agent-memory-mcp- Transport
- STDIO
- Registry
- MCP registry, active
More from kira-autonoma
Related AI & LLM Tools MCP servers
View all
Agentic Observability MCPmdfifty50-boop/agent-observability-mcpAgent tracing, cost tracking, anomaly detection for LLM agents0stars
Crawl4AI RAGcoleam00/mcp-crawl4ai-ragProvides web crawling, RAG, and knowledge-graph capabilities for AI agents with a Supabase-backed vector store.2.2Kstars
Ask Geminilykhoyda/ask-llmBridge Claude with Gemini CLI for AI-to-AI collaboration, code review, and second opinions19stars
Langchain Mcpbaixianger/langchain-mcpLangChain/LangGraph/DeepAgents knowledge + LangGraph agent debugging (Polly-like)5stars
Agent Immunedenial-web/agent-immuneAI agent security: prompt injection detection, semantic memory, output scanning, prompt hardening0stars
Authenticated Mcp Agentmikerawsonnz/gosce-agentsJWT-gated LLM gateway: authenticate (bcrypt/JWT), then run a LangChain-on-Vertex Gemini completion.0stars