python-backend-expert
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external source files provided via the
targetparameter to perform code reviews. This creates a theoretical surface for indirect prompt injection if the target files contain instructions designed to influence the agent's behavior. - Ingestion points: The
scripts/main.cjstool reads the entire content of a user-specified file usingfs.readFileSyncfor validation and analysis. - Boundary markers: The instructions do not define explicit delimiters for ingested code, although the agent is guided by specific backend rules and 'Iron Laws'.
- Capability inventory: The skill is configured with access to the
Read,Write,Edit,Bash,Grep, andGlobtools to perform its duties. - Sanitization: No sanitization or filtering is performed on the ingested source code before the agent processes it for architectural and best-practice compliance.
- [COMMAND_EXECUTION]: The 'Memory Protocol' in
SKILL.mdinstructs the agent to execute acatcommand on a local file (.claude/context/memory/learnings.md). This is a standard pattern for maintaining session context within an agent's workspace and does not involve network activity or privileged access.
Audit Metadata