mcp-registry
PassAudited by ClawScan on May 1, 2026.
Overview
This is a coherent instruction-only registry lookup skill, but users should carefully vet any third-party MCP server, remote endpoint, or credential placeholder it helps configure.
This skill appears safe for searching registry data. Before using any generated MCP configuration, verify the server source and package, inspect its requested tools and environment variables, use narrowly scoped credentials, and avoid running unknown MCP servers just to inspect them unless you trust the source.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
If a user installs a suggested MCP server, they may be adding third-party code or capabilities to their agent environment.
The skill can help generate install/connect instructions for third-party MCP packages, while the registry data itself does not describe or verify the server's actual runtime tools.
Contains: Package metadata (npm/pypi/docker identifiers), install commands, transport configuration ... Does NOT contain: Tool schemas, tool lists, or runtime behavior.
Before installing or enabling a suggested server, verify its publisher, repository, package name, documentation, requested permissions, and runtime behavior.
When users replace placeholders with real tokens, those tokens may grant account access to the configured MCP server.
The skill is expected to produce configuration snippets that may mention secrets needed by selected MCP servers, though the artifact expects placeholders rather than real credentials.
Required environment variables (like a GitHub token) should be shown as placeholder values with their descriptions.
Use least-privilege tokens, keep real secrets out of shared chats or logs, and review each server's credential requirements before use.
Connecting to an untrusted remote MCP server could expose prompts, tool requests, headers, or other interaction data depending on the selected server.
The skill may surface remote MCP endpoints and connection metadata, which can affect where agent interactions or headers are sent if the user configures them.
server.remotes[] — hosted endpoints (type, url, headers)
Only configure remote MCP endpoints you trust, and review any headers, authentication values, and data-sharing implications before connecting.