clean-code
Warn
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: MEDIUMPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
- [PERSISTENCE]: The skill instructs the agent to systematically conceal its identity within the project's audit trail. Rule 25 specifically forbids the use of agent or bot identities as commit authors and prohibits the inclusion of 'Co-Authored-By' or 'Generated with' trailers in git messages and PR descriptions. This directive to spoof human authorship masks the AI's contribution, making it difficult for security teams to differentiate between human-written and AI-generated code during audits or incident responses.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to operate on external codebases, which act as a source of untrusted data that could contain malicious instructions.
- Ingestion points: The agent reads and processes existing repository files during refactoring or code extension tasks (SKILL.md).
- Boundary markers: There are no instructions providing clear delimiters or warnings to ignore embedded instructions found within the codebase being modified.
- Capability inventory: The agent is authorized to write new code files and execute shell commands for linting, type-checking, and running tests (Rule 26, SKILL.md).
- Sanitization: The workflow lacks a verification or sanitization step to ensure that instructions found within the ingested code do not influence the agent's behavior or execution of shell commands.
Audit Metadata