incu-way-development

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local repository scripts such as run-sonar.sh and standard development tools including pnpm for type-checking, linting, and testing. It also performs Git operations to manage branches and worktrees.
  • [DATA_EXPOSURE]: Accesses configuration files located in the user's home directory (~/.ways/config.yaml) to verify repository eligibility against a blacklist and to discover linked repositories. These files are part of the incu-tech toolchain's global state.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests content from manual repository files such as PRD.md and TICKET.md. Malicious instructions embedded in these files can influence the feature requirements and implementation plans generated by the agent.
  • Ingestion points: Reads PRD.md, CLAUDE.md, and requirements from docs/requirements/{slug}/TICKET.md during Phase 0.
  • Boundary markers: None implemented; instructions from documentation are treated as functional requirements.
  • Capability inventory: File system writes, Git operations, and execution of build/test/security tools.
  • Sanitization: None; external content is interpolated directly into the process workflow.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 01:48 PM
Security Audit — agent-trust-hub — incu-way-development