| name | websocket-googleai |
| description | Update googleai_dart from Google AI WebSocket schema changes. Use for live schema refresh, change review, scaffolding, and verification. |
Google AI WebSocket Workflow
Prerequisites
- Auth: No auth env vars required for fetch/review; runtime WebSocket usage still uses Google AI credentials.
- CLI:
python3 .agents/shared/api-toolkit/scripts/api_toolkit.py
- Existing-package commands: run the repo-relative examples from the repository root. If you run them elsewhere, invoke the script via an absolute path and pass an absolute
--config-dir.
Workflow
- Fetch:
python3 .agents/shared/api-toolkit/scripts/api_toolkit.py fetch --config-dir packages/googleai_dart/.agents/skills/websocket-googleai/config
Fetch writes the candidate spec to the configured output_dir as latest-<spec>.json.
2. Review:
python3 .agents/shared/api-toolkit/scripts/api_toolkit.py review --config-dir packages/googleai_dart/.agents/skills/websocket-googleai/config
- Implement with
scaffold plus the package references, then promote the reviewed candidate from output_dir/latest-<spec>.json into packages/googleai_dart/specs/ before final verification.
- Verify:
python3 .agents/shared/api-toolkit/scripts/api_toolkit.py verify --config-dir packages/googleai_dart/.agents/skills/websocket-googleai/config --checks all --scope all
Specs
| Spec | Description |
|---|
live | Real-time bidirectional audio/video streaming via WebSocket |
Package References
Separate Dart Quality Steps
cd packages/googleai_dart
dart analyze --fatal-infos
dart format --set-exit-if-changed .
dart test test/unit/
