The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions tell the agent to advise the user to run node .gitnexus/run.cjs analyze in the terminal if the search index is stale. This involves executing a local JavaScript file within the project's hidden directory. Running unverified scripts from the workspace is a security risk as the script's content or source is not validated by the skill.
[PROMPT_INJECTION]: The skill processes external repository data and code symbols, creating a surface for indirect prompt injection attacks where malicious code could influence the agent's behavior.
Ingestion points: Targeted symbols analyzed by the impact tool, file contents from Git changes processed by detect_changes, and execution flow data read from custom gitnexus:// URIs.
Boundary markers: Absent; there are no specific instructions or delimiters provided to prevent the agent from interpreting embedded code comments or data as instructions.
Capability inventory: The skill has the capability to instruct the user to execute shell commands (via node).
Sanitization: Absent; the skill does not mention any validation, filtering, or escaping logic for the repository content it analyzes.

gitnexus-impact-analysis