The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user or agent to execute npx gitnexus analyze. This command downloads and runs code from the npm registry from an external source that is not recognized as a well-known or pre-approved vendor.
[COMMAND_EXECUTION]: The skill utilizes tools like gitnexus_rename to perform automated, multi-file edits to the codebase. While it includes a dry_run mode for safety, the execution of automated writes across a project represents a high-impact operation.
[DATA_EXPOSURE]: The skill performs deep analysis of project execution flows and dependencies (gitnexus_impact, gitnexus_context, gitnexus_cypher). While no network exfiltration was detected, it generates a detailed map of the internal project structure.
[INDIRECT_PROMPT_INJECTION]: The skill identifies an attack surface for indirect prompt injection because it ingests and processes source code (which may contain untrusted content) and possesses the capability to modify the filesystem.
Ingestion points: Reads project source code, symbol names, and execution flows via gitnexus_query and gitnexus_context.
Boundary markers: No specific delimiters or instructions to ignore embedded code comments/instructions are present.
Capability inventory: Performs file modifications through gitnexus_rename.
Sanitization: There is no evidence of sanitization for the code content being analyzed or used as the basis for automated edits.

gitnexus-refactoring