The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash and Task tools to perform verification tasks like linting, type-checking, and running unit tests. These operations are strictly restricted to read-only evidence collection, with explicit instructions prohibiting any modifications to the codebase or environment.
[EXTERNAL_DOWNLOADS]: The instructions suggest using playwright-cli for browser-based QA, providing a link to its official repository under the Microsoft GitHub organization. This reference points to a well-known service for legitimate development and testing purposes.
[PROMPT_INJECTION]: The skill manages the risk of indirect prompt injection when analyzing untrusted diffs and file content (Ingestion points: git status, Read). It implements boundary markers by requiring that results be based on 'evidence you actually collected' rather than 'what looks done'. It utilizes an 'Adversarial Subagent Review' (Capability inventory: Task, Bash) to mitigate risks by tasking independent agents with refuting claims based on specific failing cases or lines of code.

verify