profile-system-architecture
Pass
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-defined architectural profiles and evidence files through
.replacement/architecture-profile.json. - Ingestion points: The
scripts/check-profile.pyscript reads the.replacement/architecture-profile.jsonfile and all file paths specified within its evidence and adapter fields. - Boundary markers: No specific delimiters or "ignore instructions" markers are used in the prompt instructions when processing this external data.
- Capability inventory: The skill includes functions to read local files for checksum calculations and to execute git commands via subprocesses.
- Sanitization: The validator script ensures files exist and match their expected SHA256 hashes, although it does not implement strict path sandboxing for the evidence references.
- [COMMAND_EXECUTION]: The
scripts/check-profile.pyscript executes a hardcoded git command usingsubprocess.check_output()to retrieve the current repository revision. This is done using a list of arguments without enabling shell execution, which is the recommended practice for secure command invocation.
Audit Metadata