Code holds the plan
Phases, branches, Barriers, retry limits, and stopping rules live in code, where they can be reviewed and tested.
Program stability × Agent initiative
Let code drive the flow. Agents handle judgment.
Dynamic Workflow puts orchestration in TypeScript. Order, concurrency, and stopping rules become explicit; a complete Agent Loop runs only where the task needs understanding and judgment.
01 / CODE HOLDS THE PLAN
02 / AGENTS HANDLE JUDGMENT
03 / HUMANS RETAIN RESPONSIBILITY
01
WHY DYNAMIC WORKFLOW
Not every step
needs a model.
Skills carry knowledge and strategy. When mechanical steps also live in natural language, the Agent often has to interpret the plan again as it advances. Dynamic Workflow moves that control flow into code.
Agents handle judgment
agent() starts a complete ReAct Loop with tools and its own Context—not a one-shot Prompt completion.
Schema gives code a stable shape
When code consumes an Agent result, JSON Schema validates its shape before the next step runs.
02
CREATE & RUN
Create from a prompt.
Run with one command.
Generate a Workflow from one sentence, or run the bundled Deep Research example from the repository root.
Open the example
TERMINAL / CREATE
GENERATE
PROMPT → CODEX AGENT
WORKFLOW.TS → STDOUT
TERMINAL / DEEP RESEARCH
RUN
JSONL EVENTS → STDERR
RESULT → STDOUT
03
PORTABLE BY DESIGN
One file,
Run anywhere.
Run the same workflow.ts as a cloud or FaaS service, inside CI/CD, a desktop app, or directly from the CLI.
WORKFLOW.TS / PORTABLE
Bring the runtime · keep the Workflow
HOST_01
Cloud service
HOST_02
Desktop
HOST_03
CLI
HOST_04
FaaS service
HOST_05
CI/CD
04
THE WORKFLOW IS CODE
Ordinary TypeScript.
Only agent() starts an Agent Loop.
Use arrays, branches, loops, and concurrency for the mechanical work. Call agent() when a step needs understanding, exploration, or judgment.
Read the Workflow API contract
workflow.ts
01 import { agent, log, parallel, phase, pipeline } from "@deerwork-ai/deer-workflow";
02
03 export const meta = {
04 name: "research-synthesis",
05 description: "Researches and synthesizes a topic.",
06 phases: [
07 { title: "Plan" },
08 { title: "Research" },
09 { title: "Synthesize" },
10 ],
11 exampleArgs: { topic: "Agentic workflow design" },
12 };
13
14 export default async function workflow(args: { topic: string }) {
15 phase("Plan");
16 log(`Planning research for ${args.topic}`);
17 const plan = await agent<{ angles: string[] }>(
18 `Plan independent research angles for ${args.topic}`,
19 {
20 schema: {
21 type: "object",
22 properties: { angles: { type: "array", items: { type: "string" } } },
23 required: ["angles"],
24 additionalProperties: false,
25 },
26 },
27 );
28
29 phase("Research");
30 log(`Researching ${plan.angles.length} angles`);
31 const signals = await parallel(
32 plan.angles.map((angle) => () => agent(`Research ${angle}`)),
33 );
34 const findings = await pipeline(
35 signals.filter(Boolean),
36 (signal) => agent(`Extract key facts:\n${signal}`),
37 );
38 log(`Prepared ${findings.length} findings`);
39
40 phase("Synthesize");
41 return agent(`Synthesize:\n${JSON.stringify(findings)}`);
42 }
FLOW READY
EXPLICIT IMPORTS / ASYNC RUNTIME