Files
united-workforce/packages/workflow-reactor
xiaoju 26cf51366f feat: Phase 3 — engine read path + runtime context builder
- Add buildThreadContext(headHash, cas) to workflow-runtime
- Expand extract phase to return { meta, contentPayload, refs[] }
- Add parseCasThreadNode() to workflow-cas for node type parsing
- Update createWorkflow to write ContentMerkleNode with artifact refs
- Tests: 4 pass (build-context + extract-refs)
- Biome format pass on all files

Refs #155, closes #158

小橘 <xiaoju@shazhou.work>
2026-05-09 08:00:24 +00:00
..

@uncaged/workflow-reactor

LLM calling abstraction and thread “reactor” for structured tool invocation.

What This Package Does

It exposes createLlmFn (chat completion wrapper) and createThreadReactor (multi-turn tool loop configuration) plus supporting message/tool types. @uncaged/workflow-execute consumes this for extractor and supervisor paths that talk to OpenAI-style APIs with tools.

Key Exports

From src/index.ts:

  • Functions: createLlmFn, createThreadReactor
  • Types: ChatMessage, LlmFn, StructuredToolSpec, ThreadReactorConfig, ThreadReactorFn, ThreadReactorInvokeArgs, ToolCall, ToolDefinition

Dependencies

  • Workspace: @uncaged/workflow-protocol
  • Peer: zod ^4

Usage

import { createLlmFn, createThreadReactor } from "@uncaged/workflow-reactor";
// Usually composed inside @uncaged/workflow-execute rather than directly by applications.