Spur
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
Spur
Spur is an MCP-native organisational memory for software systems. It turns verified repository evidence and human decisions into typed, connected knowledge that can be explored through a web workspace or Model Context Protocol (MCP).
Spur and its generated models are in beta. Verify important answers against their cited sources.
What Spur provides
Typed knowledge for systems, integrations, information, entities, capabilities, processes, controls, decisions, rules and observations.
Commit-pinned evidence with file, line range and exact excerpt.
Candidate, accepted, disputed and stale lifecycle states, separate from confidence and authority.
Organisation-scoped PostgreSQL storage, permissions, feature switches and OAuth grants.
Repository modelling through the bundled Claude Code skill and MCP tools.
Local standard-input/output MCP and OAuth-protected remote MCP.
Optional answer synthesis through OpenAI or Anthropic; external MCP hosts can use their own model instead.
Google or passwordless email-code sign-in for shared deployments.
Related MCP server: uni-kb
Requirements
Node.js 22 LTS or Node.js 24 LTS
Git
PostgreSQL 14 or newer, or Docker with Compose
Run locally
npm install
cp .env.example .env.local
docker compose up -d postgres
npx spur migrate
npm run devOpen http://localhost:3000. The default configuration uses the bundled fixture repository, local loopback-only authentication and no built-in answer model.
To start the complete Compose stack instead:
docker compose up --buildConfigure an organisation
Set the organisation and repository in .env.local:
DATABASE_URL=postgresql://spur:spur@localhost:5432/spur
SPUR_ORGANIZATION_SLUG=example
SPUR_ORGANIZATION_NAME=Example
SPUR_REPOSITORY_ROOT=/absolute/path/to/repository
SPUR_REPOSITORY_NAME=Example application
SPUR_AUTH=none
SPUR_SYNTHESIS_PROVIDER=noneSPUR_AUTH=none is accepted only for a loopback-bound local installation. Shared deployments use Google OpenID Connect or passwordless email codes delivered through SMTP or the Mailgun adapter. See .env.example and multi-organisation setup.
Model a repository
Install the bundled Claude Code skill:
npx spur skills install claude
npx spur skills install claude --skill spur-write-requirementsConnect Claude Code to the intended Spur MCP instance and ask it to model the current committed repository with the installed spur-model-repository skill. The workflow pins one immutable commit, writes candidate knowledge with exact evidence and never promotes candidates automatically.
Use the bundled spur-write-requirements skill to draft evidence-aware PRDs or technical specifications with selectable acceptance criteria, Definition of Done, delivery sections, diagrams, images, screenshots, and wireframes. It saves an unverified draft in Spur only after explicit approval.
MCP
Run the local MCP server with the same environment as the web application:
node --env-file=.env.local bin/spur.mjs mcpShared deployments expose Streamable HTTP at /mcp and use OAuth with Proof Key for Code Exchange (PKCE). Remote MCP must be enabled for the organisation.
Validation
npm run typecheck
npm run lint
npm test
npm run buildSecurity
Report vulnerabilities according to SECURITY.md. Never commit credentials or tenant data. Spur stores session and OAuth tokens only as hashes and binds repository evidence to immutable commits.
Licence
Spur is licensed under the GNU Affero General Public License v3.0.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
Repository knowledge graph MCP server for codebase understanding and debugging.
Shared context for your codebase: a hosted code knowledge graph over MCP. Free plan, OAuth sign-up.
One MCP server for all your MCP servers: code mode, per-tool permissions, approvals and audit.
Energy-domain MCP tools for Cernion context, evidence and guarded process preparation
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceEnables AI agents to search code by meaning, explore codebase structure, store and query knowledge with temporal facts, and read source code through a set of MCP tools.7MIT
- AlicenseNot gradedqualityDmaintenanceEnables parsing, indexing, and querying source code as structured knowledge, providing code exploration, spec generation, and migration tools via 20 MCP tools.MIT
- AlicenseNot gradedqualityDmaintenanceStandalone MCP harness for cross-system process evidence, code-change impact review, and natural-language repository checkout mapping, with optional accelerators like CodeGraph.1MIT
- AlicenseNot gradedqualityBmaintenanceProvides MCP tools to search, compare, and get recommendations for open-source projects from a structured knowledge graph of GitHub data.MIT