The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill audits an entire Angular codebase (src/), which constitutes untrusted data ingestion. A malicious actor could embed instructions within comments or code in the target project to influence the agent's behavior during the audit or fix phases.
Ingestion points: The skill reads all files within the src/ directory for audit purposes.
Boundary markers: The instructions lack explicit boundary markers or warnings for the agent to ignore instructions embedded within the audited code, though it emphasizes citing the style guide.
Capability inventory: The skill can execute shell commands (pnpm, ng, rg, sed, git) and write files (reports and source code fixes).
Sanitization: No explicit sanitization of the audited content is described, but the process mandates human approval before any code modifications are performed.
[COMMAND_EXECUTION]: The skill uses several shell commands to inventory files, search for patterns, and verify code changes.
Evidence: git ls-files src | sed 's/.*\. // ' | sort | uniq -c used for file type inventory in SKILL.md.
Evidence: pnpm lint, ng lint, pnpm build, pnpm test, and pnpm exec prettier --check used for verification during the fix process in SKILL.md.
Evidence: rg used for deterministic checks against the codebase in SKILL.md.
These tools are standard for Angular development workflows and are used within their intended scope.

ng-review-style-guide