The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which could contain malicious instructions.
Ingestion points: The skill reads various files from the repository including PRDs, ADRs, existing specs, roadmaps, and source code as evidence during the grilling process.
Boundary markers: There are no explicit instructions to the agent to ignore or delimit embedded instructions within the processed documents.
Capability inventory: The skill can execute documentation checks (package scripts/CI), update local files, and return data to the calling agent.
Sanitization: The skill does not describe any sanitization or validation of the content read from external files before processing it.
[COMMAND_EXECUTION]: The skill is instructed to run documentation-related scripts and CI checks found within the target repository.
Evidence: Instructions in references/planning-artifacts.md direct the agent to "inspect package scripts and CI for documentation formatting, lint, link, and anchor checks" and "run the relevant documentation checks". It also uses git diff --check to validate changes.
[EXTERNAL_DOWNLOADS]: The skill includes a reference to an external GitHub repository for attribution purposes.
Evidence: references/resumable-planning.md contains a link to a Markdown file in Matt Pocock's GitHub repository (mattpocock/skills). This is used for workflow attribution and no content is downloaded or executed from this source.

ziw-grill