The Agent Skills Directory

[SAFE]: No significant security vulnerabilities were detected. The skill is designed for documentation workflows, reading source files and logs to generate briefs. It includes explicit warnings to exclude credentials and sensitive private data from gathered evidence.
[INDIRECT_PROMPT_INJECTION]: The skill has an ingestion surface by reading code paths and logs from the repository (SKILL.md, Step 2). These excerpts are quoted in markdown templates (assets/brief-template.md) to provide context for other LLMs. While this creates a surface for indirect prompt injection, it is mitigated by explicit instructions to exclude credentials and the lack of dangerous capabilities like network access or command execution. Ingestion points: local file system (SKILL.md). Boundary markers: absent. Capability inventory: local file writes (SKILL.md). Sanitization: credentials exclusion mentioned (SKILL.md).

to-prompt