Choose theme
AI Resources
Codebase Memory MCP
Codebase Memory MCP stores a structural map of a repository in SQLite so an AI coding agent can query it through MCP.
It is a code-analysis backend, not a coding agent. The calling agent interprets the results; the graph supplies symbols, relationships and paths. Use this as a first read, not a recommendation. Open the original project before trusting details like terms, limits, privacy, cost, setup, or safety.
What it is
Saved structure for recurring code questions
A persistent graph lets a coding agent revisit callers, imports and dependencies without starting every question with a fresh repository-wide search.
Why it stands out
A structural backend beside your existing agent
Parsing and graph queries run in the native server; it does not include an LLM. Your MCP client supplies the conversational and reasoning layer.
Availability
Local server and cross-platform releases
The repository declares MIT. It publishes macOS, Linux and Windows binaries alongside package and client-configuration paths. Local indexing does not establish how the calling agent handles returned code.
Why it matters
What makes it useful
Before changing a shared function, you may need its callers and the code paths that depend on it. Codebase Memory MCP provides a saved structural map for that investigation; the relevant implementation still needs reading.
What to know
Where it fits
Configure a supported MCP client, index the selected repository with index_repository, then ask the agent to trace inbound calls to a named function. Open the returned source paths before making a change. Reindex manually when you need current graph data and automatic updating is disabled.
Notable points
What stands out
The graph can distinguish a call relationship from a file merely mentioning the same name. The current README describes tree-sitter parsing plus additional type-resolution logic for selected languages. What can be resolved depends on the language and source; a missing graph edge is not proof that no dependency exists.
Before using
What to review
Inspect the installer and its proposed client-configuration changes. The installation path can write MCP entries and instruction blocks, not just a binary.
Choose the indexing root and protect the local graph. CBM_ALLOWED_ROOT restricts paths, but does not filter sensitive content inside an allowed repository.
Automatic indexing starts disabled in the current configuration reference. Background-watcher registration is enabled separately; neither setting describes the data handling of your calling agent.
The latest stable release is v0.11.0, published 15 September 2026. Its notes call for rebuilding an old index after the file-extension identity change. Current main-branch documentation may describe changes beyond that release.
The security policy still lists 0.10.x as supported while the release page marks v0.11.0 as latest. Check the policy and your chosen release together; the mismatch does not establish security coverage.
Reader fit
Who may find it relevant
Developers revisiting caller and dependency questions across a repository.
Teams choosing between a queryable structural graph, readable file maps and embedding-based retrieval.
Less relevant when one or two source files answer the question, or when you need a complete coding agent rather than a backend for one.
Editorial note
Why LifeHubber lists it
If you want to stop background watching, distinguish the two controls. auto_watch=false prevents later sessions from registering their own project; other watched projects can remain. For the whole watcher subsystem, the current configuration guide says to set watcher_enabled=false and stop the daemon so the next daemon reads that setting. Reconnecting the MCP client alone is not enough. Manual indexing remains available.
Source links
Source materials
Reader note
Before relying on this entry
LifeHubber lists entries to help readers inspect AI projects, not to endorse them or prove they are safe, suitable, accurate, maintained, or right for a specific use. We do not verify every entry in depth. Before relying on anything listed, review the original materials, terms, privacy practices, limits, and risks that matter for your situation.
What to explore next
Compare how the repository map is built and kept.
A persistent graph is one way to stop an agent rediscovering the same code structure. Compare it with a readable file-based map or another local SQLite-backed index before choosing the setup that fits the repository.
More in AI Agents
Keep browsing this category
Explore more AI agent projects.
Paperclip
paperclipai/paperclip
A self-hosted server and dashboard for coordinating agent teams through companies, goals, roles, issues, heartbeats, budgets, approvals, and persistent activity records.
OpenDots
CopilotKit/OpenDots
An early CopilotKit template for a persistent specialist-agent workspace with editable pages, per-Dot computers, and separately configured conversation, voice, and Slack services.
NanoClaw
nanocoai/nanoclaw
A lightweight personal agent system that runs agents in containers and connects them to messaging channels, memory, and scheduled jobs.
For project maintainers
Listed here? You can use the badge.
If you maintain a project with a current LifeHubber listing, you may add the optional “Listed on LifeHubber AI Resources” badge to its README, docs, or website. No introduction or permission request is needed.
