ProEngram
Syncs Markdown from a GitHub knowledge repository into ProEngram's searchable index and creates pull requests under docs/ for proposed Markdown changes without auto-merging.
Stores reviewed Sentry8 policy snapshots for the Policies page; direct ProEngram MCP calls are not currently evaluated by Sentry8, and live policy synchronization requires an adapter.
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.
ProEngram
ProEngram is a company knowledge dashboard and remote MCP server. Each employee connects from their own ChatGPT or Claude account through Perminister OAuth. ProEngram stores shared saved answers and indexed knowledge in Nanogres-managed PostgreSQL, while GitHub Markdown remains the canonical editable source.
Architecture
Identity and authorization: Perminister handles dashboard sign-in and issues resource-scoped user tokens for MCP. ProEngram introspects each token and checks the user's current Perminister grant for every tool action.
Shared knowledge: GitHub Markdown is synced into a searchable index in Nanogres PostgreSQL. Saved answers and citations are also shared with the user's company workspace.
Reviewable edits:
propose_markdown_changecreates a GitHub pull request underdocs/; it never merges automatically.Policies: ProEngram stores reviewed Sentry8 policy snapshots. Direct ProEngram MCP calls are not currently evaluated by Sentry8. See
docs/sentry8-integration.md.Dashboard: The company dashboard uses Panel Frame UI.
/ui-previewis a development-only gallery built with UI Preview Kit and synthetic content.
Related MCP server: MCPipe Toolchain
Local setup
Install Node.js 24.22.1 or newer in the 24.x line, plus pnpm 12.18 or newer.
Copy
.env.exampleto.env.localand fill in the settings below.Create a dedicated ProEngram database and least-privilege runtime and migration roles in Nanogres. Put the runtime TLS connection string in
DATABASE_URLand the migration-role connection string inMIGRATION_DATABASE_URL.Install dependencies with
pnpm install, apply the schema withpnpm db:migrate, then runpnpm dev. The migration command reads.env.localautomatically.Open
http://localhost:3000. Use/ui-previewfor the synthetic component gallery.
Perminister setup
Create two separate confidential ProEngram app clients in Perminister:
Dashboard OIDC client: register
http://localhost:3000/auth/perminister/callbackas the callback URL (replace the origin in production). Enable theopenid,profile,email, andperminister_apiscopes. Give the app client the dashboard access needed for sign-in andPOST /api/authorizechecks.MCP resource client: configure the exact canonical resource URI from
PERMINISTER_MCP_RESOURCE_URI(locallyhttp://localhost:3000/mcp) and the scopesbrain:read,policy:read,result:save, andknowledge:propose. Use its confidential credentials for token introspection and live action checks. Remote assistant clients register separately with Perminister through OAuth discovery.
Set PERMINISTER_ORGANIZATION_ID to the lowercase organization identifier used by this deployment, and set PERMINISTER_ISSUER to the issuer returned by Perminister metadata. Keep both client secrets and the OAuth state signing secret in the hosting environment. Use an HTTPS origin, a dedicated cookie secret of at least 32 characters, and an HTTPS resource URI in production.
The MCP endpoint is /mcp. Protected Resource Metadata is published at the RFC 9728 path for that exact resource URL (for example, /.well-known/oauth-protected-resource/mcp). ChatGPT and Claude discover Perminister there and obtain a separate user token for the ProEngram resource.
GitHub setup
Install a GitHub App on the one company knowledge repository. Grant repository Contents read/write and Pull Requests read/write permissions. Configure the app ID, installation ID, private key, owner, repository, and default branch in the environment. ProEngram syncs Markdown into Nanogres and limits proposed edits to docs/*.md or docs/*.mdx pull requests. Sync is capped at 500 Markdown files and 20 MB total; individual files above 300 KB are skipped.
Database and migrations
DATABASE_URL and MIGRATION_DATABASE_URL are server-only. Apply schema changes with pnpm db:migrate; the migration runner uses a PostgreSQL advisory lock. For hosted deployments, run migrations from a controlled release step with the migration role, then run the app with the narrower runtime role.
Sentry8 policy snapshots
The Policies page accepts a reviewed JSON catalog with policyRevision and a rules array. Import replaces the active snapshot. The Sentry8 Custom Policy API is a callback/evaluation interface, not a general active-policy export endpoint; live policy synchronization and enforcing direct MCP calls through Sentry8 require a supported identity-preserving adapter.
Guardrails
Repnix orchestrates repository health checks. Run pnpm health:audit to inventory checks, pnpm health:plan to review setup recommendations, and pnpm health to run the selected checks. Knip checks for unused code, jscpd checks duplication, Biome handles formatting and lint, and the GitHub Actions workflow runs health checks plus a production build on pull requests and pushes to main.
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
Connect your team's living knowledge base — docs, data, issues, CRM — to Claude and ChatGPT.
Your company's brain for AI agents. Cited, permission-aware knowledge across every system.
Your company brain: search decisions, people, customers and projects across connected sources.
Shared, permission-aware company context for AI agents, with provenance, approvals and audit.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceEnables AI applications to access and contextualize organizational knowledge sources including GitHub repositories and internal documentation through standardized MCP protocol integration. Features OAuth 2.1 authentication, vector-based semantic search, and optimized context chunking for enterprise development workflows.-
- FlicenseNot gradedqualityDmaintenanceProvides read-only access to Confluence, GitHub, Jira, Figma, Outlook, Teams, and browser automation for AI clients, enabling safe discovery, retrieval, and summarization of company knowledge without modifying source systems.-
- AlicenseNot gradedqualityCmaintenanceEnables AI clients like Claude, ChatGPT, and Cursor to search, read, and write back to one cross-model Markdown library of saved articles, videos, PDFs, notes, and conversations. Hosted remote server with OAuth 2.1 or Agent-key auth and read_only, edit, or full permission tiers.MIT
- AlicenseNot gradedqualityBmaintenanceEnables agents to read, search, and propose durable knowledge stored as flat files in a GitHub repository, with every write opened as a pull request and domain owners approving changes in a web UI without touching git.Apache 2.0