The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs several shell commands to manage the repository state and validate generated content.
It uses git commands including rev-parse, fetch, pull, and branch to synchronize the working environment before making modifications.
It executes generated shell scripts using bash -n for syntax validation and runs them with --help and --check flags to confirm expected behavior.\n- [DYNAMIC_EXECUTION]: The skill generates and executes code at runtime based on the project's documentation and structure.
It creates validation scripts (scripts/validate-.sh) by populating a template with logic derived from the project's runbooks.
The instructions explicitly direct the agent to run these generated scripts to verify environment state, which constitutes execution of dynamically generated content.\n- [PRIVILEGE_ESCALATION]: The skill modifies file access permissions for dynamically created assets.
It uses the chmod +x command to make generated validation scripts executable within the repository.\n- [INDIRECT_PROMPT_INJECTION]: The skill's core workflow involves processing untrusted codebase content which acts as an injection surface.
Ingestion points: It reads existing Markdown files (README.md, docs/*.md) and the entire source codebase to establish the ground truth for documentation updates.
Boundary markers: There are no defined delimiters or instructions provided to isolate processed project data from the agent's logic.
Capability inventory: The skill has broad capabilities including filesystem writes, git operations, and execution of shell scripts.
Sanitization: The skill lacks documented procedures for sanitizing or escaping the content it ingests from project files before using it to generate documentation or validation logic.

doc-manager