codeguard-go
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external Go source files and go.mod files, creating a theoretical surface for indirect prompt injection via code comments or metadata. The risk is minimized by the skill's rigid adherence to a standard workflow and a structured reporting template.\n
- Ingestion points: Local .go source files and go.mod configuration files (referenced in SKILL.md and references/tooling/toolchain.md).\n
- Boundary markers: The skill enforces a specific Output Template in SKILL.md to structure the agent's findings.\n
- Capability inventory: The skill is restricted to executing standard Go commands: go vet, gofmt, go test, and go build.\n
- Sanitization: Not explicitly mentioned; the skill relies on the agent's core safety protocols and the defined manual diff review process.
Audit Metadata