The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process and analyze untrusted external data, such as code diffs, pull request descriptions, and commit messages.
Ingestion points: The guidance in references/current-guidance.md directs the agent to read user requests, PRDs, issues, and code diffs.
Boundary markers: No specific delimiters or "ignore previous instructions" markers are defined for the content being reviewed.
Capability inventory: The skill references the ability to execute a local hook script (scripts/agent-hook.py) and access the file system.
Sanitization: There are no explicit instructions for sanitizing or escaping the content of the files or PRs being reviewed.
[COMMAND_EXECUTION]: The documentation references the execution of a local script as part of the review workflow.
Evidence: references/current-guidance.md mentions the use of scripts/agent-hook.py review to enforce review gates, which implies the agent is expected to invoke this command within the environment.

code-review