Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.truthlocks.com/llms.txt

Use this file to discover all available pages before exploring further.

Creates a new workflow definition. Workflows define a directed acyclic graph (DAG) of steps that orchestrations execute in sequence. Each step specifies its type, configuration, and dependencies — enabling complex multi-step AI pipelines with deterministic execution order. Workflows are created in draft status and must be published before they can be executed. This two-phase approach allows teams to review and validate workflow definitions before they are available to agents.

Authentication

X-API-Key
string
required
API key with workflows:write scope. Alternatively, pass a Bearer JWT token in the Authorization header.
X-Tenant-ID
string
required
Tenant identifier for multi-tenant isolation.

Request

name
string
required
Human-readable workflow name. Must be unique within the tenant namespace per version.
description
string
Description of what this workflow accomplishes.
steps
object[]
required
Array of workflow step definitions. Each step contains: - name (string, required) — Unique step identifier within the workflow - type (string, required) — Step type: llm_inference, tool_call, guardrail_check, decision, transform, webhook - config (object, required) — Step-specific configuration (model, tool name, guardrail rules, etc.) - depends_on (string[]) — Array of step names that must complete before this step executes
version
string
Semantic version of the workflow (e.g. 1.0.0). Defaults to 1.0.0 if omitted.

Response

id
string
Unique workflow identifier.
name
string
Workflow name.
description
string
Workflow description.
version
string
Workflow version.
steps
array
The validated workflow step definitions.
status
string
Workflow status. Always draft on creation.
created_at
string
ISO 8601 timestamp of creation.