The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions to run a terminal command node .gitnexus/run.cjs analyze to update a stale index. This allows the AI agent to execute a local JavaScript file within the user's workspace.- [REMOTE_CODE_EXECUTION]: By instructing the agent to run the script located at .gitnexus/run.cjs, the skill facilitates the execution of arbitrary logic defined within the repository's file structure.- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to its processing of untrusted repository data and external process definitions.
Ingestion points: Data enters the agent's context through the impact and detect_changes tools, and by reading from the gitnexus://repo/{name}/processes URI scheme.
Boundary markers: The instructions lack explicit delimiters or warnings to disregard potential instructions embedded within the code symbols or execution flows being analyzed.
Capability inventory: The skill uses terminal command execution (node), custom URI reading, and automated code analysis tools.
Sanitization: There are no documented procedures for sanitizing or validating inputs received from the workspace or the custom URI before they are processed by the agent.

gitnexus-impact-analysis