The Agent Skills Directory

[SAFE]: The skill consists entirely of instructional markdown and prompt templates designed to guide human-AI interaction. No executable scripts, binaries, or configuration files that could be abused for system access were identified.- [PROMPT_INJECTION]: The skill is designed to ingest and process external code and user-provided observations, which represents a potential indirect prompt injection surface. However, the instructions are defensive in nature.
Ingestion points: Placeholders in PROMPTS.md facilitate the ingestion of repository code and raw observations.
Boundary markers: The templates employ structured formatting, such as numbered lists for observations, which helps maintain context boundaries.
Capability inventory: The skill itself does not invoke any platform tools or perform autonomous actions; it is purely informational.
Sanitization: The provided methodology explicitly advises users to verify subagent findings manually rather than accepting them blindly, mitigating the impact of potentially poisoned data.

fresh-eyes