CreateFlowBuilderAgentOptions
Extends
Section titled “Extends”Properties
Section titled “Properties”agents?
Section titled “agents?”optional agents?: AgentConfig[];Inherited from
Section titled “Inherited from”controlModel?
Section titled “controlModel?”optional controlModel?: LanguageModel;Optional model for the control path (routing, decide, extraction), run at
temperature 0 for determinism. Defaults to model (the speaker) when unset.
Set this to pin control to a reliable provider independent of the speaker.
Inherited from
Section titled “Inherited from”description?
Section titled “description?”optional description?: string;Inherited from
Section titled “Inherited from”experimental?
Section titled “experimental?”optional experimental?: object;| Name | Type | Description |
|---|---|---|
|
|
|
Flow reply nodes: silo flow-transition control tools + deterministic evaluator.
Default ON when the agent declares |
Inherited from
Section titled “Inherited from”flows?
Section titled “flows?”optional flows?: Flow[];Inherited from
Section titled “Inherited from”globalTools?
Section titled “globalTools?”optional globalTools?: Record<string, AnyTool>;Safe, always-available tools made model-visible in EVERY speaking node turn (the agent “base layer”) — e.g. a returns/FAQ knowledge-base lookup the user might ask for mid-flow. This is an explicit allow-list: NEVER put consequential/mutating tools here (they must stay flow-gated), and they are not exposed during non-speaking collect extraction.
Inherited from
Section titled “Inherited from”guardrails?
Section titled “guardrails?”optional guardrails?: Guardrails;Inherited from
Section titled “Inherited from”handoffs?
Section titled “handoffs?”optional handoffs?: string[];Inherited from
Section titled “Inherited from”host: FlowBuilderHost;id: string;Inherited from
Section titled “Inherited from”instructions?
Section titled “instructions?”optional instructions?: Instructions;Inherited from
Section titled “Inherited from”knowledge?
Section titled “knowledge?”optional knowledge?: AgentKnowledge;Inherited from
Section titled “Inherited from”limits?
Section titled “limits?”optional limits?: Limits;Inherited from
Section titled “Inherited from”memory?
Section titled “memory?”optional memory?: AgentMemory;Inherited from
Section titled “Inherited from”model?
Section titled “model?”optional model?: LanguageModel;Inherited from
Section titled “Inherited from”optional name?: string;Inherited from
Section titled “Inherited from”policy?
Section titled “policy?”optional policy?: Policy;Decides allow / ask / deny for every tool call this agent makes. Overrides the runtime policy. This is how a delegated worker becomes genuinely read-only: the restriction is enforced at the gate, not by hoping the model respects its instructions.
Inherited from
Section titled “Inherited from”refine?
Section titled “refine?”optional refine?: RefinementCapability[];Pre-turn refinement policies. Default: none.
Inherited from
Section titled “Inherited from”routes?
Section titled “routes?”optional routes?: Route[];Inherited from
Section titled “Inherited from”routing?
Section titled “routing?”optional routing?: RoutingPolicy;Inherited from
Section titled “Inherited from”skills?
Section titled “skills?”optional skills?: SkillSource;Bundled procedural skills (Anthropic Agent Skill model): Level-1 name+description in prompt,
body via load_skill, resources via read_skill_resource. Scripts = allowedTools only.
Inherited from
Section titled “Inherited from”surfaceInstructions
Section titled “surfaceInstructions”surfaceInstructions: string;Host-specific policy composed after the shared playbook.
tools?
Section titled “tools?”optional tools?: Record<string, AnyTool>;Durable, model-callable effect tools (exactly-once on replay). Wrap raw AI SDK tools with wrapAiSdkTool().
Inherited from
Section titled “Inherited from”validate?
Section titled “validate?”optional validate?: ValidationCapability[];Post-turn validation policies (grounding/confidence gate). Default: none.
Inherited from
Section titled “Inherited from”workspace?
Section titled “workspace?”optional workspace?: AgentWorkspaceConfig;Portable workspace filesystem (or per-session resolver); auto-registers the durable workspace tool when set.
Defaults to read-only. { fs, readOnly: false } lets trusted executor code write while the
model gets a read-only traversal surface; add modelWritable: true to expose write/edit/mv/rm
to the model as well. Filesystem/mount enforcement remains authoritative.