php-exploit-chain-audit
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from external files (
vuln_audit/*.mdandauth_audit/*.md) which are generated by other sink/sub-skills during a code audit. If the project being audited contains malicious payloads (e.g., in a function name or a comment that ends up in a vulnerability report), it could influence the agent's logic during the aggregation phase. - Ingestion points: Files located in
vuln_audit/andauth_audit/directories. - Boundary markers: None specified in the instructions to protect against instructions embedded within the reports being parsed.
- Capability inventory: The skill has file-write capabilities, outputting reports to
{output_path}/exploit_chain/exploit_chains_{timestamp}.md. - Sanitization: There is no explicit instruction to sanitize or escape content from input reports before interpolating them into the final aggregation template.
Audit Metadata