Interface: BuildLlmCallerOptions

Defined in: packages/agentos/src/orchestration/planning/buildLlmCaller.ts:26

Options for building an LLM caller function.

At minimum, provide provider OR model (or both). If neither is provided, auto-detection from env vars kicks in.

Properties

apiKey?

optional apiKey: string

Defined in: packages/agentos/src/orchestration/planning/buildLlmCaller.ts:32

API key override (not needed for CLI providers).

---

baseUrl?

optional baseUrl: string

Defined in: packages/agentos/src/orchestration/planning/buildLlmCaller.ts:34

Base URL override (e.g. for OpenRouter, Ollama).

---

maxTokens?

optional maxTokens: number

Defined in: packages/agentos/src/orchestration/planning/buildLlmCaller.ts:38

Max tokens for planning calls. Default: 4096.

---

model?

optional model: string

Defined in: packages/agentos/src/orchestration/planning/buildLlmCaller.ts:30

Model ID: 'gpt-4o', 'claude-opus-4-6', 'gemini-2.5-flash', etc.

---

provider?

optional provider: string

Defined in: packages/agentos/src/orchestration/planning/buildLlmCaller.ts:28

Provider ID: 'openai', 'anthropic', 'claude-code-cli', 'gemini-cli', etc.

---

temperature?

optional temperature: number

Defined in: packages/agentos/src/orchestration/planning/buildLlmCaller.ts:36

Temperature for planning calls. Default: 0.3.