Claude Skill MartBrowse skillsWhat is a Skill?
 
← Back to list

Agent Designer (Multi-Agent Architecture)

Designs multi-agent architectures from requirements, generates validated Anthropic/OpenAI tool schemas, and evaluates execution logs for cost, latency, and bottlenecks.

Dev & CodingAdvanced★ 24,151⑂ 3,405AI score 8/10Last updated: Aug 9, 2026

What it does

agent-designer replaces freehand architecture sketching with three deterministic scripts.

  • Architecture planning: agent_planner.py scores your requirements JSON (goal, tasks, response-time/budget/concurrency constraints, team size) and selects among Single, Supervisor, Pipeline, Hierarchical, and Swarm patterns — returning agent roles, communication links, a Mermaid diagram, and an implementation roadmap.
  • Tool schema generation: tool_schema_generator.py turns plain-language tool descriptions into provider-ready schemas for both Anthropic and OpenAI, with a hard gate that every tool must report ✓ Valid.
  • Execution evaluation: agent_evaluator.py produces success rate, latency distribution, error analysis, cost breakdown, SLA compliance, and ranked optimization recommendations.
  • Verification loop: the design isn't finished until there are 0 invalid schemas and 0 critical issues on a pilot run.

Who it's for

  • AI engineers building or refactoring multi-agent systems
  • Teams that must keep function-calling schemas consistent across model providers
  • Tech leads who need hard data on where an agent pipeline burns time and money

Examples

  1. Research automation: "Design an agent architecture for research automation" → planner returns a Supervisor pattern with specialist agents plus a Mermaid diagram to share with the team.
  2. Schema batch: Feed descriptions of ten internal APIs and get tools_anthropic.json / tools_openai.json with a validation summary before wiring anything up.
  3. Bottleneck triage: Run a pilot's logs through the evaluator, find that the retrieval agent dominates p95 latency, apply the top recommendation, and re-measure.

· · · Install guide · · ·

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the .gemini/skills/agent-designer folder from the GitHub repo alirezarezvani/claude-skills into my ~/.claude/skills/agent-designer/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/alirezarezvani/claude-skills.git && mkdir -p ~/.claude/skills && cp -r claude-skills/.gemini/skills/agent-designer ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

  1. Open a terminal and confirm Python 3 is available: python3 --version
  2. Clone the repository: git clone https://github.com/alirezarezvani/claude-skills.git
  3. Create the skills directory: mkdir -p ~/.claude/skills
  4. Copy the entire skill folder (the scripts, assets/, and references/ are all required): cp -r claude-skills/.gemini/skills/agent-designer ~/.claude/skills/
  5. Restart Claude Code and trigger it with a request like "design a multi-agent architecture for our support workflow".
  6. For your first run, copy assets/sample_system_requirements.json and edit the values rather than writing the JSON from scratch.
View source on GitHub ↗License: MIT