The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local system commands through its included Python script and explicit agent instructions.
The scripts/scan_docs.py script utilizes subprocess.run to interact with Git for identifying documentation drift and changed files.
Instructions in SKILL.md (Step 5) direct the agent to execute documentation build tools (such as mkdocs, docusaurus, or sphinx) and run sample commands or code snippets found within the target repository's documentation.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests and processes untrusted documentation content from external repositories.
Ingestion points: The agent is tasked with reading the content of all documentation files identified by scripts/scan_docs.py to reconcile them against the codebase.
Boundary markers: The instructions do not define explicit delimiters or security warnings to prevent the agent from obeying instructions embedded within the documentation files it processes.
Capability inventory: The skill possesses shell execution capabilities through subprocess.run (in scripts/scan_docs.py) and instructions to run arbitrary code snippets and build tools discovered in the processed data.
Sanitization: There is no evidence of sanitization or validation of the documentation content before the agent processes it and attempts to verify it through execution.

docs-update-expert