The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data such as PRDs, architecture documentation, system context, and API responses during the Plan and Execute phases. This creates a surface where malicious instructions embedded in these documents could attempt to influence the agent's behavior.
Ingestion points: SKILL.md (Phase 1, Phase 3), references/test-plan-template.md (Section 5).
Boundary markers: None explicitly defined to delimit external requirement text from agent instructions.
Capability inventory: The skill has access to Bash, Read, Grep, and Glob tools as defined in the YAML frontmatter.
Sanitization: No explicit sanitization or filtering of external requirement content is mentioned.
[COMMAND_EXECUTION]: The skill is configured to use the Bash tool, which is necessary for its primary purpose of executing test suites (unit, integration, functional) and verifying fixes as described in the workflow and templates.

qa-engineer