The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes output from external test frameworks and runners (Go testing, vitest, playwright, pytest, Makefile, Taskfile). If these outputs contain malicious instructions from an attacker, the agent might attempt to follow them.
Ingestion points: Captures and reports output from shell-executed test runners and frameworks (e.g., make test, npm test, go test).
Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are used when processing tool output.
Capability inventory: Executes shell commands via project runners and writes reports to the local file system (e.g., docs/qa/).
Sanitization: No explicit sanitization or validation of the ingested tool output is mentioned before reporting or decision-making.
[COMMAND_EXECUTION]: The skill's primary function is to discover and execute shell commands via project-specific wrappers like make, task, or npm, or direct calls to go test, vitest, and pytest. It executes these commands based on the content of local configuration files (Makefile, package.json).

z-qa-orchestrator