The Agent Skills Directory

[SAFE]: The skill uses well-known diagnostic tools and libraries, such as Lighthouse and the Next.js bundle analyzer, which are sourced from established and trusted providers.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from lighthouse.json, which is generated from external web content audits. This constitutes a potential indirect prompt injection surface if the audited website were to contain malicious payloads designed to influence the agent via the tool output.
Ingestion points: lighthouse.json generated by npx lighthouse in SKILL.md.
Boundary markers: None present for the processed JSON output.
Capability inventory: Shell execution (npx, bun, cat, jq) and file system access as described in SKILL.md.
Sanitization: None present; uses jq for specific field extraction which limits the surface for instruction injection.

frontend-performance