← all repositories
deerwork-ai/deer-workflow

TypeScript calls the shots; agents fill in the blanks

An open-source runtime that keeps AI workflow orchestration in readable TypeScript so you can review what your agents are about to do instead of trusting a black-box conversation.

deer-workflow
Collecting fresh signals — velocity needs a few days of history.
collecting data…
star history

What it does

Deer Workflow generates and executes agent graphs from natural language prompts. You describe a workflow—say, researching topics in parallel and synthesizing a report—and it emits a TypeScript module that defines the control flow, phases, and failure handling. The semantic work inside each node is delegated to replaceable coding agents such as OpenAI Codex, Claude Code, or Pi.

The interesting bit

The project treats TypeScript as the source of truth for orchestration rather than hiding the plan inside an opaque multi-turn agent conversation. That means diffs, code review, and version control apply to your execution graph. It also streams structured JSONL events for automation while offering a phase-aware TUI when you want to watch agents work in real time.

Key highlights

  • Vendor-neutral agent interface: swap Codex for Claude or Pi without rewriting the graph.
  • Execution is observable via an interactive TUI or machine-readable JSONL event streams.
  • Ships with example workflows for deep research and blog writing.
  • MIT licensed and positioned as the pilot open-source core for ByteDance’s DeerFlow 3.0.

Verdict

Worth a look if you want agent orchestration that lives in version control rather than chat logs. Skip it if you need a mature, point-and-click workflow builder.

Frequently asked

What is deerwork-ai/deer-workflow?
An open-source runtime that keeps AI workflow orchestration in readable TypeScript so you can review what your agents are about to do instead of trusting a black-box conversation.
Is deer-workflow open source?
Yes — deerwork-ai/deer-workflow is open source, released under the MIT license.
What language is deer-workflow written in?
deerwork-ai/deer-workflow is primarily written in TypeScript.
How popular is deer-workflow?
deerwork-ai/deer-workflow has 501 stars on GitHub.
Where can I find deer-workflow?
deerwork-ai/deer-workflow is on GitHub at https://github.com/deerwork-ai/deer-workflow.

heatdrop uses Google Analytics to see which pages get read — nothing else. Your call. How we handle data.