build

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection as its execution flow is governed by external data in the form of implementation plans.
  • Ingestion points: Processes files from docs/plan/ or paths provided via user arguments.
  • Boundary markers: The skill does not implement delimiters or safety warnings to distinguish between the plan's data and the agent's instructions.
  • Capability inventory: Possesses filesystem write access, shell command execution (Bash), and the ability to create pull requests.
  • Sanitization: No sanitization or verification steps are present for the instructions contained within the ingested plans.
  • [EXTERNAL_DOWNLOADS]: Instructions mandate the installation of dependencies if they are referenced in the plan but missing from the environment, which could result in the installation of untrusted third-party code.
  • [COMMAND_EXECUTION]: The skill uses shell commands for development workflows, including file system navigation, executing test suites, running linters, and cleaning up temporary review reports.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 09:06 PM
Security Audit — agent-trust-hub — build