The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests potentially untrusted data from the repository, such as dependency manifests, test configurations, and existing source files, to detect the project's testing dialect. This creates a surface for indirect prompt injection where malicious instructions hidden in the codebase could be processed by the agent.
Ingestion points: Instructions specify reading the project runner, dependency manifest, test configuration, and existing tests in SKILL.md.
Boundary markers: The instructions do not define specific delimiters or guardrails to differentiate between codebase data and instructional content.
Capability inventory: The skill is capable of writing new test files and executing local build or collection tools.
Sanitization: There are no requirements for sanitizing or escaping the content read from the repository files before it is processed by the agent.
[COMMAND_EXECUTION]: The instructions direct the agent to run local shell commands to verify the generated test stubs.
Evidence: The skill specifies using go build, go test -c, and pytest collection to ensure the generated artifacts are compilable and correctly identified by the project's test runner.

z-qa-spec-writer