Vibeleaderboard
Index / agent
Visit github.com
Category
AI Agents
Rank
No. 1720Tools index

Previous survey · No. 1703 ·

Pricing
Open Source
Type
AGENT
GitHub
9 stars
Date

About

Multi-agent orchestration patterns built on top of the Claude Code CLI.

What it does

A runnable weather demonstration that fans one request into 195 country agents, waits for every result, writes a combined temperature report, then asks a final agent to calculate the average. A single slash command starts the workflow, while a remote weather service supplies each capital’s reading.

Why it's ranked here

The project makes parallel fan-out and ordered fan-in unusually concrete. Its command spells out when agents run together and when later stages must wait. That clarity makes it useful for learning orchestration, although the weather-specific design and large hand-maintained agent list limit reuse without substantial editing.

What's good

Each worker has a narrow contract: fetch one capital’s temperature and return only a Celsius value. The writer and calculator remain separate, so collection, persistence, and aggregation have visible boundaries. Saved example reports show the intended artifacts, including how an unavailable service appears among otherwise numeric results.

Tradeoffs

The workflow manually enumerates 195 workers and uses a distinct weather tool for every country, creating considerable configuration repetition. It depends on one remote HTTP service. The example data includes an unavailable reading, but the documentation does not explain whether averages exclude missing values, retry failures, or stop when a worker fails.

How to use it well

Use it as a worked example when designing jobs with many independent fetches followed by ordered aggregation. It best suits Claude Code users willing to adapt agent prompts, service tools, and output stages. It does not provide a general scheduler, a reusable orchestration library, or a documented policy for retries and partial failure.

Technical notes+

.claude/commands/orchestrate.md explicitly enumerates 195 background Task launches, waits through TaskOutput, then runs writer and average agents sequentially. .mcp.json configures one remote HTTP MCP endpoint. The sampled country definitions in .claude/agents/weather-fetch/agent-weather-chad.md, .claude/agents/weather-fetch/agent-weather-cuba.md, .claude/agents/weather-fetch/agent-weather-fiji.md, and .claude/agents/weather-fetch/agent-weather-iran.md each bind one capital to one MCP tool and constrain output to Celsius. .claude/settings.json enables all project MCP servers, selects bypassPermissions, and registers asynchronous lifecycle hooks. .codex/hooks.json separately registers five Codex hook events. output/temperatures.md contains one unavailable-service entry, while output/average.md contains a numeric result.

Observed

Primary format
Markdown agent and command definitions with JSON configuration
Launch interface
Claude Code slash command
External interface
Remote HTTP MCP server with country-specific weather tools
Workflow structure
195 parallel fetch agents, followed by sequential writer and average agents
Hook runtime
Python 3
Hook platform support
Windows, macOS, and Linux are documented
Output surface
Markdown reports for country temperatures and the calculated average

Read from README.md, .mcp.json, CLAUDE.md, .codex/hooks.json, output/average.md, .claude/settings.json, .vscode/settings.json, output/temperatures.md, .codex/hooks/HOOKS-README.md, .claude/hooks/HOOKS-README.md, .claude/commands/orchestrate.md, .claude/agents/weather-fetch/agent-weather-chad.md, .claude/agents/weather-fetch/agent-weather-cuba.md, .claude/agents/weather-fetch/agent-weather-fiji.md, .claude/agents/weather-fetch/agent-weather-iran.md.

What it can do

  • Orchestrate multiple AI agents to work together on complex tasks

    Task requirements and agent configurationCoordinated multi-agent workflow execution

  • Generate code using distributed agent collaboration

    Code requirements and specificationsGenerated code files and modules

  • Execute parallel code analysis tasks across multiple agents

    Source code files and analysis parametersComprehensive code analysis results

  • Coordinate agent communication and task delegation

    Agent roles and communication protocolsStructured agent interaction patterns

  • Manage agent workflow pipelines and dependencies

    Workflow definitions and dependency mappingsExecuted pipeline with status tracking

  • Scale code processing tasks across multiple Claude instances

    Large codebases and processing requirementsDistributed processing results

Tags

claude-codemulti-agentorchestrationcli

Tech Stack

Python

Comments (0)

No comments yet

Editorially curated, with community endorsements as a secondary signal. Corrections welcome.