The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a local Python script (scripts/score.py) for deterministic scoring and uses system tools like curl and npx for data retrieval and browser automation.
[EXTERNAL_DOWNLOADS]: The skill suggests using npx playwright screenshot to capture evidence. This involves downloading and running the Playwright tool from the NPM registry, which is a well-known service used for legitimate web automation.
[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection because it processes untrusted data from live URLs or provided code samples.
Ingestion points: Target URLs, screenshots, and code paths provided by the user or fetched from the web as defined in SKILL.md.
Boundary markers: The instructions do not define specific delimiters or 'ignore' warnings for the external content being evaluated.
Capability inventory: The skill can perform network requests (curl), execute shell commands (npx, python3), and write reports to the local filesystem (./SLOP-REPORT.md).
Sanitization: No explicit sanitization or validation of the ingested external content is mentioned before it is processed by the agent.

slop-eval