The Agent Skills Directory

[SAFE]: The skill implements a standard developer workflow for test debugging. It uses recognized testing tools and follows security best practices such as least privilege and PII redaction.
[INDIRECT_PROMPT_INJECTION]: The skill accepts user-provided arguments (file paths and test names) to execute Playwright commands. While this presents a theoretical attack surface, it is a core functional requirement of the debugging skill and is mitigated by the global security protocols described in the instructions.
Ingestion points: User-provided $ARGUMENTS (test file path, name, and description) in SKILL.md.
Boundary markers: None explicitly surrounding the variable interpolation.
Capability inventory: Execution of terminal commands via npx playwright test and docker as specified in SKILL.md.
Sanitization: The global protocols instruct the agent to treat external data as hostile and redact secrets before sharing context.

fix