The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes GitNexus CLI commands through a local runner located at .gitnexus/run.cjs. This includes high-level operations such as analyze, status, clean, wiki, and list which interact with the repository's file system and external APIs.
[EXTERNAL_DOWNLOADS]: The instructions facilitate the download of the gitnexus package and its dependencies (like @ladybugdb/core and tree-sitter) from the npm registry using npx or pnpm dlx. This is standard behavior for Node.js-based developer tools.
[DATA_EXFILTRATION]: The wiki command includes a --gist flag designed to publish repository documentation to a public GitHub Gist. While this is an intended feature of the tool, it represents a potential channel for data exposure if sensitive information is included in the generated wiki.
[CREDENTIALS_UNSAFE]: The skill handles LLM API keys for documentation generation. It notes that keys are saved to ~/.gitnexus/config.json and can be provided via the --api-key flag. This is a standard practice for CLI tools requiring external API access.
[PROMPT_INJECTION]: The analyze command parses all source files in a repository to build a knowledge graph. This ingestion of untrusted codebase data represents a surface for indirect prompt injection, where malicious instructions in comments or code could theoretically influence the agent's behavior during subsequent analysis steps.

gitnexus-cli