The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch and analyze user-generated reviews from the Google Play Store, creating a surface for indirect prompt injection attacks.
Ingestion points: Review data is ingested into the agent's context through commands described in SKILL.md (Sections 6 and 7) and references/review-management.md, such as gpc reviews list and gpc reviews analyze.
Boundary markers: There are no instructions or delimiters provided to help the agent distinguish between its core instructions and the potentially untrusted content found within user reviews.
Capability inventory: The skill possesses significant capabilities that could be targeted by an injection, including the ability to reply to reviews (gpc reviews reply), promote app versions (gpc releases promote), and halt active rollouts (gpc watch --on-breach halt).
Sanitization: No sanitization or filtering logic is mentioned for the external review content beyond a character length check for replies.
[COMMAND_EXECUTION]: The file scripts/detect_gpc.mjs is an executable script that uses execSync to run shell commands (gpc --version, gpc auth status, etc.) on the local system to verify the environment configuration.
[EXTERNAL_DOWNLOADS]: Documentation in references/ci-gating.md recommends the installation of the @gpc-cli/cli package from the NPM registry, which is the necessary tool for the skill's operation.

gpc-vitals-monitoring