polish
Pass
Audited by Gen Agent Trust Hub on Mar 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed to autonomously run build and test suites for the target project during its self-healing validation phase. It also chains execution to multiple other skills (/ux, /scale-audit, /qa, /audit) using the Agent and Task tools, which results in significant command execution across the local environment.
- [DATA_EXFILTRATION]: The skill attempts to record execution metadata into a hidden directory in the user's home folder (
~/.claude/projects/skill-telemetry.md). While described as telemetry, writing data to hidden configuration paths outside the project root is a sensitive operation. - [PROMPT_INJECTION]: The skill possesses a large indirect prompt injection surface because it is instructed to ingest and process the entire application codebase without boundary markers or sanitization. Malicious instructions embedded in the codebase could influence the agent's autonomous 'fix' behavior or the command execution during the build/test phases.
- Ingestion points: Processes the entire codebase or specific modules passed via
$ARGUMENTS. - Boundary markers: None present; the agent treats the codebase content as data to be evaluated and acted upon.
- Capability inventory: Includes file modification, skill chaining, and arbitrary shell command execution (build/test).
- Sanitization: None present.
Audit Metadata