The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read and trust content from .agents/qa-project-context.md to skip discovery questions, which creates a surface for indirect prompt injection if that file is attacker-controlled.
Ingestion points: SKILL.md instructions to check .agents/qa-project-context.md first.
Boundary markers: Absent; no instructions provided to ignore or escape embedded instructions within the context file.
Capability inventory: The skill leverages shell commands including curl and sentry-cli for health checks and issue monitoring.
Sanitization: Absent; the agent is expected to use the project context as-is.
[COMMAND_EXECUTION]: The skill contains multiple executable command snippets for the user or agent to run during verification, including:
curl commands for health checks and latency measurement.
sentry-cli commands for querying recent application issues.
jq for parsing health check JSON responses.

release-readiness