The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external test cases (from documentation, UI, or manual specs) to generate executable visual test scripts. This represents a potential indirect prompt injection surface where malicious instructions in the input could attempt to influence the generated code. However, the risk is mitigated by the skill's explicit focus on generation rather than execution, and its inclusion of safety guidelines.
Ingestion points: Test cases sourced from qa-testcase-from-docs, qa-testcase-from-ui, or manual specifications as described in the SKILL.md workflow.
Boundary markers: None explicitly defined in the provided instruction text to separate untrusted test case data from the agent's instructions.
Capability inventory: The skill has the capability to write generated test scripts to the tests/visual directory as specified in the YAML frontmatter.
Sanitization: The skill includes instructions to mask sensitive content such as timestamps, avatars, and PII in the SKILL.md and references/best-practices.md files to prevent data exposure in screenshots.
[SAFE]: No malicious patterns or behaviors were detected. The skill uses well-known industry tools (Playwright, Percy, BackstopJS) for its primary purpose and emphasizes secure practices, such as storing tokens in environment variables and avoiding the capture of secrets in screenshots.

qa-visual-regression-writer