The Agent Skills Directory

[SAFE]: The skill is designed to improve repository documentation standards. It performs local repository audits and scaffolds documents based on project type detection (e.g., Go, Python, Node.js).
[COMMAND_EXECUTION]: The skill utilizes local bash scripts (audit-oss-docs.sh) to verify the existence of documentation files and project manifests. These operations are restricted to the local file system and are necessary for the skill's primary function of documentation auditing.
[DATA_EXPOSURE]: The audit script reads basic project metadata such as the repository name and git origin URL to provide context in its report. This data is processed locally and not transmitted to external servers.
[PROMPT_INJECTION]: The skill includes reference patterns for AGENTS.md which instruct AI assistants on session completion tasks (e.g., committing and pushing changes). These are standard development workflows intended to ensure project hygiene and are not malicious overrides.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests data from untrusted project manifests (like pyproject.toml or package.json) to determine project type and generate documentation templates.
Ingestion points: scripts/audit-oss-docs.sh and references/project-types.md read file names and content from the repository root.
Boundary markers: None identified in the provided scripts or templates.
Capability inventory: The skill uses Bash, Write, Edit, Glob, and Grep tools.
Sanitization: No explicit sanitization of manifest content is performed, though the impact is limited to the content of generated documentation files.

oss-docs