cs-roadmap

Pass

Audited by Gen Agent Trust Hub on Apr 27, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local Python scripts (search-yaml.py, validate-yaml.py) and standard utilities like grep within the project's codestable/ directory. These tools are used for administrative tasks such as searching existing documentation and validating the YAML syntax of generated roadmap items.
  • [PROMPT_INJECTION]: The skill processes untrusted user input and external documentation to generate technical plans, which constitutes an indirect prompt injection surface.
  • Ingestion points: User-provided materials (Phase 2), existing files in codestable/requirements/, and codestable/architecture/ directories.
  • Boundary markers: The skill does not explicitly use delimiters for untrusted input, but it implements a mandatory human-in-the-loop review (Phase 5) before finalizing outputs.
  • Capability inventory: File system write access to codestable/roadmap/ and execution of local shell commands via Python.
  • Sanitization: There is no explicit logic described to sanitize user-provided keywords used in shell command interpolation (e.g., {大需求关键词}).
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 27, 2026, 02:59 AM