flutter-app-workflow
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The 'Preflight' and 'Understanding Gate' instructions (references/project-preflight.md) require the agent to reconcile external instructions with the codebase and confirm scope with the user, protecting against malicious or unclear directions.
- [SAFE]: The skill includes explicit instructions in 'Git handoff' (references/delivery-verification.md) to exclude credentials, temporary probes, and local skill artifacts from the repository.
- [SAFE]: Development is restricted to standard Flutter and Dart tools, and the skill explicitly forbids the unauthorized or silent installation of external packages or additional agent skills.
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external GitHub tickets and issues, creating a potential vector for indirect instructions.
- Ingestion points: GitHub ticket titles, bodies, and comments (references/project-preflight.md).
- Boundary markers: No technical delimiters are used to separate ticket data from agent instructions.
- Capability inventory: The agent can perform file modifications and execute shell commands via Git and Flutter CLI tools.
- Sanitization: The risk is mitigated through mandatory human clarification steps for any ambiguity and a structured post-implementation review process.
Audit Metadata