systematic-debugging
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external technical data which creates a surface for indirect prompt injection.
- Ingestion points: Error messages, stack traces, logs, and git diffs (SKILL.md Phase 1).
- Boundary markers: Absent.
- Capability inventory: Read, Grep, Glob, and Bash (SKILL.md allowed-tools).
- Sanitization: Absent.
- [EXTERNAL_DOWNLOADS]: The skill contains references to an external GitHub repository for methodology attribution.
- Evidence: Links to github.com/obra/superpowers/ found in SKILL.md, defense-in-depth.md, and root-cause-tracing.md.
Audit Metadata