Repo Memory
A git-backed memory layer that lets AI agents share what they learn about your codebase across sessions and tools. Exposes five MCP tools: get_repo_memory pulls all prior context at session start, add_fact records verified claims with file locations and evidence, list_facts filters by tag or source file, add_decision captures architectural choices, and add_gotcha logs surprises. Everything writes to a .ai-memory/ directory you commit like code, so facts sync through normal git push/pull. No database, no SaaS, works offline. Reach for it when multiple agents or developers hit the same codebase and you're tired of re-explaining where the auth middleware lives or why you picked httpx over requests.
- GitHub stars
- 1
- tools
- 5
- transport
- STDIO
- env var
- 1
- last updated
- Aug 26, 2026
Tools
5 toolsPublic tool metadata for what this MCP can expose to an agent.
get_repo_memoryRead the repo's stored memory; call at the start of any task on this repo.
Read the repo's stored memory; call at the start of any task on this repo.
No parameter schema in public metadata yet.
add_factRecord a verified non-obvious fact (location, behavior, convention), with evidence so the next agent can re-verify it.
Record a verified non-obvious fact (location, behavior, convention), with evidence so the next agent can re-verify it.
No parameter schema in public metadata yet.
list_factsList only the facts in a specific area, filtered by tag or source file.
List only the facts in a specific area, filtered by tag or source file.
No parameter schema in public metadata yet.
add_decisionRecord a non-trivial choice (architecture, library, trade-off) and why it was made.
Record a non-trivial choice (architecture, library, trade-off) and why it was made.
No parameter schema in public metadata yet.
add_gotchaRecord a surprise that wasted time as a one-line gotcha note.
Record a surprise that wasted time as a one-line gotcha note.
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
REPO_MEMORY_ROOTAbsolute path to the repository whose .ai-memory/ to read/write. Defaults to the current working directory.
Readme
repo-memory
Shared, git-tracked working memory for AI agents that share a codebase. What one Claude / Cursor / Cline learns about your repo, the next one picks up automatically. No database. No SaaS. Just files in your repo.
The problem
Every AI session that touches your repo starts from zero. It re-greps the same files. It re-discovers the same conventions. It re-asks the same questions you already answered three sessions ago. Multi-user / multi-tool makes it worse: your teammate's Cursor and your Claude Code learn the same codebase independently.
Details
- Repository
- github.com/yubinkim444/repo-memory
- Categories
- AI & LLM Tools, Developer Tools
- Package
repo-memory-mcp- Transport
- STDIO
- Registry
- MCP registry, active
More from yubinkim444
Related AI & LLM Tools MCP servers
View all
Knowledge Graph Mcpzcsabbagh/knowledge-graph-mcpKnowledge graph MCP for student learning with spaced repetition and mastery tracking1stars
Agent Analyticsagent-analytics/agent-analytics-mcpAnalytics your AI agent can actually use. Track, experiment, and optimize via MCP.0stars
A2atlassianagentic-eng/a2atlassianJira & Confluence for AI agents. Pre-configured connections, compact output.0stars
i18n-agenti18n-agent/mcp-clientAI-powered translation for 48 languages with context-aware quality0stars
Free Agent Toolsakiba-labs/free-agent-tools43 free no-auth tools: Japan life, visas, travel and work checks, plus pitch roast.0stars
Agent Brokerbasilalshukaili/agentbroker23 MCP tools: compliance, verification, messaging, booking, US contracts. 15 need no key.0stars