Ravenroot Second Brain MCP
OfficialClick on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Ravenroot Second Brain MCP what are the god nodes in my second brain?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Ravenroot Second Brain MCP
Run the Ravenroot knowledge graph as a local Model Context Protocol server. The repository contains a read-only graph snapshot generated from ravenroot-ai/ravenroot, the FastMCP server, and the code that rebuilds the graph. It does not require Google Cloud, Horizon, Qwen, or access to the source repository to answer queries.
Start locally
Install uv and clone this repository. On macOS or Linux:
git clone https://github.com/ravenroot-ai/ravenroot-second-brain.git
cd ravenroot-second-brain
./start-mcp.shOn Windows, run ./start-mcp.ps1 from PowerShell after cloning. Both scripts perform the same verification and start the stdio server.
Use the absolute path to start-mcp.sh as the command for an MCP client that launches local stdio servers. The first start installs the pinned Python dependencies, rebuilds second_brain/composed/graph.json, verifies its SHA-256 against second_brain/provenance.json, and starts the server. Subsequent starts reuse the verified graph. Python 3.12 or 3.13 is selected by uv.
For a client on the same machine that requires an HTTP URL, run:
MCP_TRANSPORT=http MCP_HOST=127.0.0.1 MCP_PORT=8765 ./start-mcp.shThe endpoint is http://127.0.0.1:8765/mcp. It has no authentication; keep it bound to loopback unless you add access controls yourself. This repository does not host a shared public endpoint. Each clone runs its own server.
Related MCP server: intelligraph-mini
Snapshot and updates
second_brain/parts/graph_partN.json is the committed distribution format. GitHub blocks ordinary Git files above 100 MiB, so the single rebuilt graph is ignored. second_brain/graph_parts.py compose joins the parts in order and refuses a result whose SHA-256 differs from the split header. second_brain/verify_bundle.py also checks the published provenance. The graph and provenance identify the indexed Ravenroot source commit; they are an immutable snapshot until this repository publishes a new one.
The private graph generation workflow and canonical source graph stay outside this repository. Maintainers publish a newly accepted snapshot by updating only the parts and provenance together, then running the tests from a clean checkout before pushing. Never commit second_brain/graph.json, provider settings, credentials, local caches, or agent configuration.
MCP tools
The read-only server provides graphify_query, graphify_path, graphify_explain, graphify_neighbors, graphify_stats, and graphify_god_nodes, plus graphify://stats and graphify://provenance resources.
Development
uv sync --frozen --dev
uv run pytest -q second_brainThe code and bundled data are distributed under Apache License 2.0.
Available Tools
10 toolsget_communityB
Get all nodes in a community by community ID.
Inputs: community_id (required), project_path, token_budget.
get_neighborsC
Get all direct neighbors of a node with edge details.
Inputs: label (required), project_path, token_budget, relation_filter.
get_nodeB
Get full details for a specific node by label or ID.
Inputs: label (required), project_path.
get_pr_impactA
Get detailed graph impact for a specific PR: which files it changes, which knowledge-graph communities are affected, and how many nodes are touched. Use this to assess merge risk or check for overlap with your current work.
Inputs: pr_number (required), repo, project_path.
god_nodesB
Return the most connected nodes - the core abstractions of the knowledge graph.
Inputs: top_n, project_path.
graph_statsA
Return summary statistics: node count, edge count, communities, confidence breakdown.
Inputs: project_path.
list_prsA
List open GitHub PRs with CI status, review state, and graph impact (which communities each PR touches, blast radius). Use this before starting work to check if a PR already covers the area you're about to change.
Inputs: base, repo, project_path.
query_graphB
Search the knowledge graph using BFS or DFS. Returns relevant nodes and edges as text context.
Inputs: question (required), mode, depth, project_path, +2 more.
shortest_pathA
Find the shortest path between two concepts in the knowledge graph. Follows stored edge direction by default; set undirected=true to ignore it.
Inputs: source (required), target (required), max_hops, undirected, +1 more.
triage_prsA
Return all actionable open PRs (correct base, not stale) with full graph impact data so you can reason about review priority, merge order, and conflict risk. Call this when the user asks 'what PRs should I review?' or 'what's ready to merge?'
Inputs: base, repo, project_path.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 10 tools
The graph tools (query_graph, get_node, get_neighbors, get_community, god_nodes, graph_stats, shortest_path) each target a clearly distinct operation. The PR tools are mostly distinct, though list_prs and triage_prs overlap heavily—both return open PRs with graph-impact data—and only the descriptions' filtering nuance (actionable/correct base/not stale) separates them.
All names use consistent snake_case, which is predictable and readable. However, the verb pattern is not strictly uniform: several are noun phrases (god_nodes, graph_stats, shortest_path) while others are verb_noun (get_node, list_prs), a minor deviation.
Ten tools is well-scoped for a knowledge-graph query plus PR-triage server, with each tool earning its place. No redundancy that would push it toward bloat.
Read-side graph coverage is solid (search, fetch, neighbors, community, hubs, stats, pathfinding) and the PR workflow is covered from listing to impact to triage. The only gap is the absence of any mutation tools (add/update/delete node or edge), which an agent might expect for a 'second brain'.
Maintenance
Related MCP Connectors
Repository knowledge graph MCP server for codebase understanding and debugging.
Query your own knowledge graph from any MCP client. The graph lives in a database you own.
Personal knowledge graph as an AI memory layer over MCP - read, save, and link your memories.
MCP server for querying BrainKB, a knowledge base for neuroscience knowledge graphs.
Related MCP Servers
- FlicenseNot gradedqualityNot gradedmaintenanceA Knowledge Graph MCP server optimized for LLM context efficiency through compact JSON and SQLite persistence. It enables full graph management including node/edge CRUD operations, full-text search, and subgraph traversal.-
- FlicenseAqualityDmaintenanceLocal-first MCP graph intelligence server providing RRF hybrid search, multi-hop traversal, source snippets, and rationale nodes for AI agents, without Docker or web UI.5-
- AlicenseAqualityDmaintenancePersistent knowledge graph MCP server with SQLite backend. Enables graph traversal, fuzzy search, temporal queries, and timestamps for entity management.12MIT
- FlicenseNot gradedqualityDmaintenanceEnables local, deterministic semantic operations on OWL/RDF ontologies, including Description Logic reasoning, SPARQL querying, TBox/ABox mutation, and interactive graph visualization via an MCP interface.1-