AgentAlley
Submit

Agent workflow designer

Design production-grade multi-agent workflows with clear pattern choice (sequential, parallel, hierarchical), handoff contracts, failure handling, and cost/context controls.

How to use it

Claude Code
  1. Run the line below. It pulls the whole folder into ~/.claude/skills/agent-workflow-designer.
  2. Describe your job in plain words. Claude Code follows the skill from there.
Claude Code — installs the whole folder, not just SKILL.md
npx degit alirezarezvani/claude-skills/engineering/skills/agent-workflow-designer#main ~/.claude/skills/agent-workflow-designer

For one project only, change the path to .claude/skills/agent-workflow-designer.

Claude (web or desktop app)
  1. On this page open ⋯ → Download .md.
  2. Save it as SKILL.md in a folder, zip the folder, then Customize → Skills → + → Create skill → Upload a skill.
  3. Pick the file and Save. Claude shows the name and description and runs a security scan.
  4. Check the skill is switched on.
  5. Start a new chat and describe your job in plain words. The AI follows the skill from there.
ChatGPT or another app
  1. ChatGPT: make a Project and paste it into Instructions.
  2. Neither? Paste it at the top of a new chat — it works for that chat.
Not working?
  • Check which app you pasted it into — the steps above name the right one.
  • Some skills need the paid tier of Claude or ChatGPT.
Step-by-step guide with screenshots · Ask in the forum

Paste into Claude, ChatGPT or Cursor.

Source of Agent workflow designer

Show the full text84 lines
agent-workflow-designer/SKILL.md84 lines · 2.7 KB
Outline
RawView on GitHub
namedescription
agent-workflow-designerDesign production-grade multi-agent workflows with clear pattern choice (sequential, parallel, hierarchical), handoff contracts, failure handling, and cost/context controls. Use when architecting a multi-step agent pipeline, choosing between single-agent vs multi-agent approaches, or refactoring an LLM workflow that suffers from context bloat or unreliable handoffs.

Agent Workflow Designer

Tier: POWERFUL
Category: Engineering
Domain: Multi-Agent Systems / AI Orchestration

Overview

Design production-grade multi-agent workflows with clear pattern choice, handoff contracts, failure handling, and cost/context controls.

Core Capabilities

  • Workflow pattern selection for multi-step agent systems
  • Skeleton config generation for fast workflow bootstrapping
  • Context and cost discipline across long-running flows
  • Error recovery and retry strategy scaffolding
  • Documentation pointers for operational pattern tradeoffs

When to Use

  • A single prompt is insufficient for task complexity
  • You need specialist agents with explicit boundaries
  • You want deterministic workflow structure before implementation
  • You need validation loops for quality or safety gates

Quick Start

# Generate a sequential workflow skeleton
python3 scripts/workflow_scaffolder.py sequential --name content-pipeline

# Generate an orchestrator workflow and save it
python3 scripts/workflow_scaffolder.py orchestrator --name incident-triage --output workflows/incident-triage.json

Pattern Map

  • sequential: strict step-by-step dependency chain
  • parallel: fan-out/fan-in for independent subtasks
  • router: dispatch by intent/type with fallback
  • orchestrator: planner coordinates specialists with dependencies
  • evaluator: generator + quality gate loop

Detailed templates: references/workflow-patterns.md

  1. Select pattern based on dependency shape and risk profile.
  2. Scaffold config via scripts/workflow_scaffolder.py.
  3. Define handoff contract fields for every edge.
  4. Add retry/timeouts and output validation gates.
  5. Dry-run with small context budgets before scaling.

Common Pitfalls

  • Over-orchestrating tasks solvable by one well-structured prompt
  • Missing timeout/retry policies for external-model calls
  • Passing full upstream context instead of targeted artifacts
  • Ignoring per-step cost accumulation

Best Practices

  1. Start with the smallest pattern that can satisfy requirements.
  2. Keep handoff payloads explicit and bounded.
  3. Validate intermediate outputs before fan-in synthesis.
  4. Enforce budget and timeout limits in every step.
1---
2name: "agent-workflow-designer"
3description: "Design production-grade multi-agent workflows with clear pattern choice (sequential, parallel, hierarchical), handoff contracts, failure handling, and cost/context controls. Use when architecting a multi-step agent pipeline, choosing between single-agent vs multi-agent approaches, or refactoring an LLM workflow that suffers from context bloat or unreliable handoffs."
4---
5 
6# Agent Workflow Designer
7 
8**Tier:** POWERFUL
9**Category:** Engineering
10**Domain:** Multi-Agent Systems / AI Orchestration
11 
12---
13 
14## Overview
15 
16Design production-grade multi-agent workflows with clear pattern choice, handoff contracts, failure handling, and cost/context controls.
17 
18## Core Capabilities
19 
20- Workflow pattern selection for multi-step agent systems
21- Skeleton config generation for fast workflow bootstrapping
22- Context and cost discipline across long-running flows
23- Error recovery and retry strategy scaffolding
24- Documentation pointers for operational pattern tradeoffs
25 
26---
27 
28## When to Use
29 
30- A single prompt is insufficient for task complexity
31- You need specialist agents with explicit boundaries
32- You want deterministic workflow structure before implementation
33- You need validation loops for quality or safety gates
34 
35---
36 
37## Quick Start
38 
39```bash
40# Generate a sequential workflow skeleton
41python3 scripts/workflow_scaffolder.py sequential --name content-pipeline
42 
43# Generate an orchestrator workflow and save it
44python3 scripts/workflow_scaffolder.py orchestrator --name incident-triage --output workflows/incident-triage.json
45```
46 
47---
48 
49## Pattern Map
50 
51- `sequential`: strict step-by-step dependency chain
52- `parallel`: fan-out/fan-in for independent subtasks
53- `router`: dispatch by intent/type with fallback
54- `orchestrator`: planner coordinates specialists with dependencies
55- `evaluator`: generator + quality gate loop
56 
57Detailed templates: `references/workflow-patterns.md`
58 
59---
60 
61## Recommended Workflow
62 
631. Select pattern based on dependency shape and risk profile.
642. Scaffold config via `scripts/workflow_scaffolder.py`.
653. Define handoff contract fields for every edge.
664. Add retry/timeouts and output validation gates.
675. Dry-run with small context budgets before scaling.
68 
69---
70 
71## Common Pitfalls
72 
73- Over-orchestrating tasks solvable by one well-structured prompt
74- Missing timeout/retry policies for external-model calls
75- Passing full upstream context instead of targeted artifacts
76- Ignoring per-step cost accumulation
77 
78## Best Practices
79 
801. Start with the smallest pattern that can satisfy requirements.
812. Keep handoff payloads explicit and bounded.
823. Validate intermediate outputs before fan-in synthesis.
834. Enforce budget and timeout limits in every step.
84 

Discussion

Alternatives

Also in Workflow automationSee all 58 in Operations →
Browser Automation SkillWeb browser automation with AI-optimized snapshots for claude-flow agentsCoding · MITWeb Extract — Structured Data from the Open WebExtract structured JSON from web pages, search engines, and entire sites in ONE call — {title, summary, sections, key_metrics, outgoing_links, author, date, page_type, ...} fields, no second LLM pass to parse HTML. Six endpoints: scrape (single URL), scrape-interactive (JS-rendered pages with click/scroll/type), search (Google SERP + deep-scrape), map (URL discovery), crawl + crawl-status (async recursive crawl). Markdown/raw HTML on request. USE when the user needs page DATA — product pricing/specs, article fields, link graphs, JS-heavy SPAs, Google results with content. Prefer over browser-act (automation/screenshots) and WebFetch (static, no JS, no structured fields). Not for citation-rich research (use deep-research). Trigger (EN): scrape this URL, extract data from page, crawl this site, deep-scrape search results, map a domain's URLs, render this JS page. 触发词：抓取/爬取/网页提取/结构化抽取/搜索带内容/全站爬取/JS 渲染抓取/点击后抓取. Requires ZOODATA_API_KEY (free key: https://zoodata.ai/en/api-keys).Sales & ecommerce · MITAI workflow automation specialistAct as an AI Workflow Automation Specialist, guiding users in automating business processes, optimizing workflows, and integrating AI tools effectively.Infrastructure & ops · CC0-1.0Cyber security character workflowThis is a structured image generation workflow for creating cyber security characters. The workflow includes steps such as facial identity mapping, tactical equipment outfitting, cybernetic enhancements, and environmental integration to produce high-quality, cinematic renders. After uploading your face and filling in the values in the fields, your prompt is ready. NOTE: The sample image belongs to me and my brand; unauthorized use of the sample image is prohibited.Creator · CC0-1.0

Browse more free Claude skills or everything in Operations.