Architecture Impact
Pass
Audited by Gen Agent Trust Hub on Mar 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell commands and git operations to validate the environment and determine the scope of code changes. These commands are used to extract repository names, check git status, and manage the documentation cache.
- [EXTERNAL_DOWNLOADS]: The
ensure_arch_repo.shscript downloads the architecture repository from the official GitHub account of the vendor. This is a standard mechanism to keep the reference documentation up to date in the local cache. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and analyzes code changes (git diffs) which could contain malicious instructions embedded in comments or code. These instructions could potentially influence the behavior of the subagent during analysis.
- Ingestion points: Git diffs and repository file content.
- Boundary markers: No explicit boundary markers or 'ignore' instructions are used when passing code content to the analysis subagent.
- Capability inventory: The skill utilizes a subagent to read files and provide architectural recommendations.
- Sanitization: There is no evidence of sanitization or filtering of the code content before it is processed by the LLM.
Audit Metadata