lighthouse-bridge
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from repository files and Lighthouse reports which could contain malicious instructions. \n
- Ingestion points: The skill is designed to search and read
.github/workflows/*.ymlfiles, Lighthouse configuration files (e.g.,lighthouserc.json), and report data from workflow artifacts or temporary public storage as defined in the capabilities section ofSKILL.md. \n - Boundary markers: Boundary markers are absent; the instructions do not specify any delimiters or safety prompts to isolate external content from the agent's logic. \n
- Capability inventory: Based on
SKILL.mdandagents/openai.yaml, the skill is restricted to read-only access. It lacks subprocess calls, exec/eval, file-writing capabilities, and network operations (except for reading reports), and model invocation is disabled via frontmatter. \n - Sanitization: Sanitization is absent; while the skill extracts specific keys from the JSON reports, it does not mention validating or escaping the strings within those fields.
Audit Metadata