Subgraph Registry

Discover and filter 15,500+ The Graph subgraphs by domain, network, protocol type, or natural language goal with reliability scores and query URLs.

This skill appears to be what it says: a pre-built, read-only registry served locally. Before installing, consider the following: 1) The server auto-downloads a registry.db from the skill's GitHub raw URL on first run — if you want to inspect the DB first, clone the repo and review the file or run the server in an isolated/container environment. 2) The package includes Python crawlers that, if you run them, will contact The Graph gateway and can use a GATEWAY_API_KEY from a .env file — only provide credentials if you intend to rebuild/update the registry. 3) The MCP/HTTP server listens on a local port (default 3848) and exposes an SSE endpoint; only enable --http/--http-only if you trust the local network environment. 4) Verify the GitHub repository/release source if provenance matters to you. If you need additional assurance, run the npx command inside a sandbox (container/VM) and/or inspect src/index.js and the downloaded registry.db before allowing remote agents to connect.

Type: OpenClaw Skill Name: subgraph-registry-mcp Version: 0.3.2 The subgraph-registry-mcp bundle is a legitimate toolset for discovering and classifying subgraphs on The Graph Network. It includes a crawler (crawler.py), a rule-based classification engine (classifier.py), and an MCP server (src/index.js and mcp_server.py) that provides tools for searching and recommending subgraphs based on natural language goals. The server downloads a pre-built SQLite database from the project's GitHub repository (PaulieB14/subgraph-registry) to provide offline search capabilities. No evidence of data exfiltration, malicious execution, or harmful prompt injection was found; the code logic is transparent and aligns with its stated purpose of facilitating blockchain data discovery.