The Agent Skills Directory

[COMMAND_EXECUTION]: The documentation instructs the agent to execute a local Node.js script to update the repository index: node .gitnexus/run.cjs analyze. This represents a vector for executing code within the user's local environment based on instructions from the skill.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of data from a codebase into the agent's context, which could allow malicious instructions embedded in the code (e.g., in comments or strings) to influence agent behavior.
Ingestion points: Tools like query, context, impact, trace, cypher, and explain return data derived directly from the analyzed repository.
Boundary markers: There are no documented delimiters or instructions to ignore potentially malicious content within the tool outputs.
Capability inventory: The skill provides capabilities for repository navigation, data querying, and command execution through a local script.
Sanitization: The guide does not specify any sanitization or filtering processes to neutralize embedded instructions within the codebase analysis results.

gitnexus-guide