adr-audit
Pass
Audited by Gen Agent Trust Hub on Jul 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes 'fitness functions' (automated architecture checks) that are either documented within the project's ADR files or discovered via common patterns in the repository (e.g., ArchUnit, dependency-cruiser). This is the primary mechanism for auditing architecture compliance and is limited to project-defined verification scripts.
- [EXTERNAL_DOWNLOADS]: The skill can identify missing architectural dependencies mentioned in the project's rules and offer to install them using standard package managers like npm, pip, maven, or cargo. These operations target well-known package registries and require explicit user consent before proceeding.
- [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it ingests and interprets rules from the repository's markdown files (ADRs and AGENTS.md) to drive its auditing logic.
- Ingestion points: docs/adr/*.md, AGENTS.md (processed via ls, find, and grep)
- Boundary markers: Absent (the skill treats the content as normative instructions)
- Capability inventory: Shell command execution, package installation, and file creation within the docs/adr/audits/ directory
- Sanitization: Absent (relies on the user's control over the repository content)
Audit Metadata