flutter-frontend-for-adk
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches framework documentation from
https://adk.dev/llms.txtand clones a sample repository fromhttps://github.com/google/adk-samples.git. These resources are used to align the frontend implementation with the Agent Development Kit (ADK) specifications. - [COMMAND_EXECUTION]: The skill instructs the agent to perform multiple system-level operations:
- Scaffolding the Flutter application using the
flutterCLI tool. - Validating the generated code through
flutter format,flutter analyze, andflutter test. - Modifying existing project configuration and build scripts, including
Makefile,pyproject.toml,package.json,run.sh, andbuild.sh, to integrate the new frontend into the developer workflow. - [INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes local Python source code (e.g.,
app/agent.pyand ADK framework files in the virtual environment) to identify API endpoints and state schemas. This constitutes a vulnerability surface if the analyzed code contains malicious instructions. - Ingestion points: Local Python agent code and framework definitions in the workspace.
- Boundary markers: The instructions do not define explicit delimiters or 'ignore' instructions for the agent when documenting the discovered interface in
AGENT_INTERFACE_NOTES.md. - Capability inventory: The skill has the capability to write files to the workspace, modify executable build scripts, and run shell commands via the Flutter and Dart SDKs.
- Sanitization: No sanitization or validation of the ingested code content is specified in the discovery phases.
Audit Metadata