sec-triage
Pass
Audited by Gen Agent Trust Hub on Nov 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes untrusted data from external security scanners and issue trackers.
- Ingestion points: The skill reads external findings from SARIF files, GitHub Code Scanning alerts, Dependabot alerts, and tickets (
SKILL.md, Input section). - Boundary markers: Absent. There are no instructions to use delimiters or specific ignore-tags when the agent processes the descriptions or evidence from these untrusted sources.
- Capability inventory: The skill executes local scripts (
sec-scan/scripts/secfind.py,ticket_writer.py) and interacts with GitHub to create issues or private advisories (SKILL.md, Steps 3, 7, 9). - Sanitization: While Step 2 mentions masking secret values, there is no mention of sanitizing or escaping the natural language content within findings before the agent evaluates them against the security policy.
Audit Metadata