The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The mcp.json file configures the agent to download the gitnexus package from the public NPM registry using npx -y gitnexus@latest. This occurs every time the MCP server is initialized.
[REMOTE_CODE_EXECUTION]: By invoking npx, the skill executes code downloaded from a remote repository on the host system. The use of the @latest tag ensures the most recent version is used, which could lead to unexpected behavior if the package is updated or compromised.
[COMMAND_EXECUTION]: The skill defines a Model Context Protocol (MCP) server that runs a shell command (npx) to execute the GitNexus tool. The SKILL.md file also encourages users to run npx gitnexus analyze manually in their terminal.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface (Category 8) because it ingests and processes untrusted data from local code repositories and execution flow maps (gitnexus://repo/{name}/processes).
Ingestion points: Data enters the context via gitnexus_impact results and repository execution flows.
Boundary markers: No specific delimiters or instructions to ignore embedded commands in the analyzed code are provided.
Capability inventory: The skill executes subprocesses via the defined MCP server.
Sanitization: There is no evidence of sanitization or filtering of the content analyzed from the target repository before it is presented to the agent.

gitnexus-impact-analysis