workflow-integration-test-green

Pass

Audited by Gen Agent Trust Hub on Aug 13, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements a structured software engineering workflow that relies on internal tool invocations (e.g., $integration-test-verify, $debug-investigate, $spec) to manage testing cycles. All activities are scoped to the project's own codebase and documentation.
  • [COMMAND_EXECUTION]: The workflow involves executing local integration tests via $integration-test-verify. This is an expected and intended capability for a development-focused AI agent skill designed to drive test suites to a green state.
  • [DATA_EXPOSURE]: The skill accesses project-specific configuration files (docs/project-config.json), reference documentation (docs/project-reference/), and maintains iteration logs in plans/reports/. This data access is necessary for providing context-aware development assistance and does not involve sensitive system or user credentials.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from repository documentation and specifications.
  • Ingestion points: Files under docs/project-config.json, docs/project-reference/, and docs/specs/.
  • Boundary markers: Not explicitly defined for document ingestion, though structured workflows are used to process them.
  • Capability inventory: The skill can write code via $fix and execute code via test runners.
  • Sanitization: Not explicitly mentioned, but the skill operates within a controlled development environment where the agent's actions are subject to user review and task tracking.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 13, 2026, 03:11 PM
Security Audit — agent-trust-hub — workflow-integration-test-green