The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches and executes the 'gitnexus' package from the npm registry using npx or pnpm dlx if the local runner is absent.
[REMOTE_CODE_EXECUTION]: Instructs the agent to execute code directly from the npm registry via npx gitnexus analyze and pnpm dlx gitnexus@latest analyze.
[COMMAND_EXECUTION]: Executes a dynamically generated script node .gitnexus/run.cjs that is created in the project directory during the analysis process.
[PROMPT_INJECTION]:
Ingestion points: Repository source files parsed by the analyze and wiki commands (referenced in SKILL.md).
Boundary markers: None identified to prevent malicious instructions within the source code from influencing documentation output.
Capability inventory: Generates repository documentation (wiki) and context files (CLAUDE.md/AGENTS.md) using an LLM based on source code content.
Sanitization: No evidence of sanitization or filtering of source code content before it is processed by the LLM.
[DATA_EXFILTRATION]: The wiki --gist flag publishes generated repository documentation to a public GitHub Gist, which could lead to accidental exposure of sensitive project information.
[CREDENTIALS_UNSAFE]: Directs the agent to handle and store LLM API keys in ~/.gitnexus/config.json for documentation generation tasks.

gitnexus-cli