# Gemini Structured Output Rules
Last reviewed: 2025-08-11
Use schema-constrained JSON with Gemini for reliable parsing and evaluation.
## Rules
- **Always set** `responseMimeType: "application/json"` and provide a **minimal** `responseSchema`.
- **Keep schemas small**: Schema tokens count against input budget; avoid overly deep nesting and huge enums.
- **Validate** responses with `zod` and handle parse failures with a bounded retry.
- **Batch** where feasible using `generateBatch()` with identical `generationConfig` + caching.
- **Token hygiene**: Use `countTokens()` to budget prompts and consider pre-trimming with `adaptivePrompt()`.
## Example (TypeScript)
```ts
const response = await model.generateContent({
contents: [{ role: 'user', parts: [{ text: prompt }] }],
generationConfig: {
responseMimeType: 'application/json',
responseSchema: {
type: 'object',
properties: {
summary: { type: 'string' },
sources: { type: 'array', items: { type: 'string' } }
},
required: ['summary', 'sources']
},
// optional tuning
temperature: 0.4,
topP: 0.9,
topK: 40,
candidateCount: 1,
maxOutputTokens: Number(process.env.GEMINI_MAX_OUTPUT_TOKENS || 65536)
}
});
```
## Project Conventions
- **Schemas**:
- `outline`: `{ outline: string[] }`
- `sections`: `{ sections: string[], citations?: string[] }`
- `summary`: `{ summary: string }`
- **Env flags**: `ENABLE_GEMINI_GOOGLE_SEARCH`, `ENABLE_GEMINI_CODE_EXECUTION`, `ENABLE_GEMINI_FUNCTIONS`, `ENABLE_PROVIDER_CACHE`.
## References
- Structured Output: https://ai.google.dev/gemini-api/docs/structured-output
- Vertex AI notes on `responseMimeType`/schema: https://cloud.google.com/vertex-ai/generative-ai/docs/multimodal/control-generated-output
