Singularity - MCP Discovery & Security Gateway

Singularity — MCP Discovery & Security Gateway

Perplexity for MCP. Discover the right tool for a problem, see its trust card (prompt-injection risk, permission overreach, publisher status), connect it in one click, and execute it through a secure, audited gateway — one product, one config.

See ROADMAP.md for the full plan. Blueprint: ../Singularity_Blueprint.md.

Quickstart (local)

pip install -r requirements.txtcopy .env.example .env          # optional, defaults are fine (SQLite)python scripts/seed_index.py    # builds index + embeddings (first run downloads ~35MB model)python -m uvicorn app.main:app --reload   # http://127.0.0.1:8000/docs

A fresh boot with an empty database auto-seeds itself — seed_index.py is only needed for manual re-seeding.

Use Singularity as an MCP server

Singularity is a native MCP server (streamable HTTP, stateless) at /mcp.

VS Code / Copilot Chat — .vscode/mcp.json:

{ "servers": { "singularity": { "type": "http", "url": "http://127.0.0.1:8000/mcp" } } }

Cursor — ~/.cursor/mcp.json:

{ "mcpServers": { "singularity": { "url": "http://127.0.0.1:8000/mcp" } } }

Claude Code CLI:

claude mcp add --transport http singularity http://127.0.0.1:8000/mcp

Then just ask your agent: "my team needs expense tracking" — it calls find_solutions and gets ranked options with trust cards and security flags.

One-click install (replace the URL after deploying): Install in VS Code

/mcp now requires OAuth (a popup at /authorize/<slug> supplies the access token). For a curl-able check use /health and /.well-known/oauth-authorization-server.

Tools

Every result carries fit_score, trust_score, rank_score, trust_flags, and a human-readable rationale. The seed index includes quickledger-pro, an intentionally poisoned fictional tool — ask about expenses and watch it get flagged (suspicious_description_imperative, hidden_unicode_characters, permission_overreach) and pushed down the ranking.

Deploy your own instance (Render + Supabase)

Full walkthrough in DEPLOY.md. Summary:

1. Create a free Supabase project → copy the connection string → add +psycopg driver segment + ?sslmode=require
2. Push this repo to GitHub → Render "New +" → "Blueprint" → it reads render.yaml
3. Paste the Supabase URL into the SINGULARITY_DATABASE_URL env var when prompted
4. First boot auto-creates tables and seeds the index; https://<app>.onrender.com/mcp is live

Publish to MCP registries

After deploying (public HTTPS required), list Singularity so others can find it:

1. Official MCP Registry — edit placeholders in server.json (replace XXXX with your Render app name), then:
   powershell

   npx mcp-publisher@latest init      # scaffolds/validates server.jsonnpx mcp-publisher@latest login     # prove io.github.<you> ownership via GitHubnpx mcp-publisher@latest publish   # listed at https://registry.modelcontextprotocol.io within minutes

2. Smithery — submit repo at https://smithery.ai/new (uses smithery.yaml)
3. Glama — submit remote URL at https://glama.ai, then claim the listing
4. GitHub topics — add mcp-server, mcp, ai-security topics to your repo; PulseMCP/MCP.so crawlers pick it up within 1–2 weeks

Tip: registries re-crawl weekly and strip verified badges from sleeping servers — Render's $7/mo Starter plan keeps Singularity always-on during listing/investor periods.

Tests

pytest -v

Layout

singularity/  main.py                     FastAPI app factory + lifespan (auto-seed, mounts MCP)  mcp_server.py               native MCP server: find_solutions / get_trust_report / compare_tools  config.py                   env-driven settings (SINGULARITY_* prefix)  database.py                 SQLAlchemy engine/session (SQLite dev, Postgres prod)  models.py                   Tool + AuditLog tables  schemas.py                  request/response contracts  middleware/    rate_limit.py             per-IP token-bucket limiter (X-Forwarded-For aware)  services/    discovery_engine.py       heuristic intent parsing    recommendation_index.py   in-memory vector search over seeded tools    ranking.py                fit/trust blend + rationale builder    scanner.py                static injection/overreach scan (Phase 2 preview)    seeder.py                 seed logic shared by CLI and boot-time auto-seed    embeddings.py             fastembed provider (bge-small-en-v1.5)  routers/    recommend.py              POST /recommend, GET /toolsdata/seed_tools.json          curated 21-tool index incl. demo attack sample (tier1/tier2/tier3)scripts/seed_index.py         manual seeding CLIscripts/smoke_execute.ps1     authenticated REST execute smoke testscripts/smoke_oauth.py        OAuth discovery + config smoke testDockerfile / render.yaml      one-command Render deploymentserver.json / smithery.yaml   registry publishing manifests (fill placeholders first)tests/                        pytest suite (API + scanner + MCP tools & endpoint)