| name | openapi-googleai |
| description | Update googleai_dart from Google AI OpenAPI changes. Use for spec refresh, change review, scaffolding, and verification. |
Google AI OpenAPI Workflow
Prerequisites
- Auth:
GEMINI_API_KEY, GOOGLE_AI_API_KEY
- 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 the spec you are updating:
python3 .agents/shared/api-toolkit/scripts/api_toolkit.py fetch \
--config-dir packages/googleai_dart/.agents/skills/openapi-googleai/config \
--spec-name main
Fetch writes the candidate spec to the configured output_dir as latest-<spec>.json.
2. Review the same spec:
python3 .agents/shared/api-toolkit/scripts/api_toolkit.py review \
--config-dir packages/googleai_dart/.agents/skills/openapi-googleai/config \
--spec-name main
- Implement with
scaffold plus the package references, then promote the candidate spec. The verifier reads specs_dir/{local_file} — copy using the exact local_file name from config/specs.json, not the spec-name:
cp /tmp/openapi-googleai-dart/latest-main.json packages/googleai_dart/specs/openapi.json
cp /tmp/openapi-googleai-dart/latest-interactions.json packages/googleai_dart/specs/openapi-interactions.json
- Verify:
python3 .agents/shared/api-toolkit/scripts/api_toolkit.py verify \
--config-dir packages/googleai_dart/.agents/skills/openapi-googleai/config \
--spec-name main \
--checks all --scope all
Specs
| Spec | Description |
|---|
main | Core Gemini API - generation, embeddings, files, models, etc. |
interactions | Server-side state, agents, background execution |
Use --spec-name interactions when reviewing or verifying the interactions spec.
Package References
Separate Dart Quality Steps
cd packages/googleai_dart
dart analyze --fatal-infos
dart format --set-exit-if-changed .
dart test test/unit/
