flutter-ci-cd

Pass

Audited by Gen Agent Trust Hub on Sep 5, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill instructions demonstrate a strong security posture by explicitly prohibiting the printing, movement, or commitment of signing material and service credentials.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process repository configuration files which represents a standard data ingestion surface.
  • Ingestion points: Reads pubspec.yaml, lockfiles, SDK configuration, and existing pipeline definitions (SKILL.md, references/provider-and-monorepo.md).
  • Boundary markers: The skill explicitly mandates separating pull-request validation from release authority and protected environment access (SKILL.md, references/delivery-and-secrets.md).
  • Capability inventory: The skill orchestrates local repository commands and CI provider configurations but restricts these to existing delivery models without allowing arbitrary execution.
  • Sanitization: Instructions require validating provider syntax with official linters and performing dry-run validations before any mutation (SKILL.md, references/provider-and-monorepo.md).
  • [DATA_EXPOSURE_AND_EXFILTRATION]: No evidence of credential exposure or data exfiltration. The skill recommends using provider-native secret management and short-lived credentials (references/delivery-and-secrets.md).
  • [REMOTE_CODE_EXECUTION]: The skill does not perform remote script execution or download untrusted code. It recommends pinning third-party pipeline components to immutable references (references/delivery-and-secrets.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 5, 2026, 07:30 AM
Security Audit — agent-trust-hub — flutter-ci-cd