security-audit
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted project source code as its primary task, creating a surface for indirect prompt injection if the source code contains malicious instructions targeting the auditor agent.
- Ingestion points: Target project source code files accessed during reconnaissance, hunting, and validation phases (SKILL.md).
- Boundary markers: The instructions mandate the use of
sandbox-runtimefor enforced isolation when performing local validation of vulnerability candidates (SKILL.md, Section 8). - Capability inventory: The skill possesses capabilities to trace data flows and potentially execute local build or test processes within a sandbox to verify findings.
- Sanitization: The skill relies on architectural isolation rather than specific string sanitization of the analyzed content.
- [EXTERNAL_DOWNLOADS]: The skill references a repository from Cloudflare as a design inspiration.
- Evidence: Mentions the Cloudflare security-audit-skill repository at
github.com/cloudflare/security-audit-skill(SKILL.md, Section 17). - Note: This reference targets a well-known technology organization.
Audit Metadata