affaan-m/claude-swarm — an AI tool — sits at 406 GitHub stars in the AI Tools space. Multi-agent orchestration for Claude Code — decompose tasks, coordinate agents, visualize everything in a rich terminal UI
Snapshot summary built from the project's own GitHub metadata — there's no written TopGit review yet. The page will update automatically when a full review is published.
WHY NO REVIEW YET
TopGit writes full reviews for the most-starred, most-requested repositories. This page is a snapshot until then — see the READ ME tab for the original README in full.
Multi-agent orchestration for Claude Code — decompose complex tasks into parallel subtasks, coordinate agents in real-time, and visualize everything in a rich terminal UI.
Built with the Claude Agent SDK for the Claude Code Hackathon (Feb 10-16, 2026).
How It Works
You: "Refactor auth module from Express middleware to Next.js API routes"
Claude Swarm:
Phase 1: Opus 4.6 decomposes task into dependency graph
Phase 2: Parallel agents execute subtasks with live dashboard
Phase 2.5: Opus 4.6 Quality Gate reviews all agent outputs
Phase 3: Results summary with costs and session replay
Task Decomposition — Describe a complex task. Opus 4.6 analyzes your codebase and breaks it into a dependency graph of subtasks
Parallel Agent Spawning — Independent subtasks run simultaneously via Claude Agent SDK. Dependent tasks wait.
Real-time Coordination — File conflict detection prevents agents from stepping on each other. Budget enforcement stops runaway costs.
Opus Quality Gate — After agents complete, Opus 4.6 reviews the combined output for correctness, consistency, and completeness
Rich Terminal UI — htop-style dashboard showing agent progress, tool usage, costs, and file conflicts in real-time
Session Replay — Every swarm execution is recorded. Replay any session to review what each agent did.
Quick Start
# See it in action instantly (no API key needed!)
pip install claude-swarm
claude-swarm --demo
# Or install from source
git clone https://github.com/affaan-m/claude-swarm
cd claude-swarm
pip install -e .
# Set your API key for real usage
export ANTHROPIC_API_KEY="sk-ant-..."
# Run a swarm
claude-swarm "Refactor auth module from Express middleware to Next.js API routes"
# Dry run (shows plan without executing)
claude-swarm --dry-run "Add user authentication with JWT"
# Custom budget, agents, and retries
claude-swarm --budget 3.0 --max-agents 6 --retry 2 "Build a REST API for user management"
# Disable quality gate for faster execution
claude-swarm --no-quality-gate "Quick fix: update README"
Pessimistic file locking prevents agents from editing the same file simultaneously
Budget enforcement
Hard cost limit — cancels remaining tasks when budget is exceeded
Cost tracking
Real-time per-agent and total cost monitoring
Opus Quality Gate
Phase 2.5 — Opus 4.6 reviews all agent outputs for correctness and consistency
Smart model selection
Opus 4.6 for planning + quality review, Haiku for worker agents (3x cheaper)
Task retry
Failed tasks are automatically retried with configurable attempt limits
Demo mode
--demo flag shows animated TUI without API key (great for presentations)
Session recording
Every swarm execution recorded as JSONL events
Session replay
claude-swarm replay <id> to review what each agent did
YAML config
Declarative swarm topologies via swarm.yaml
Progress visualization
Live htop-style dashboard with Rich
CLI Reference
# Main command
claude-swarm [OPTIONS] TASK
Options:
-d, --cwd TEXT Working directory (default: .)
-n, --max-agents INTEGER Max concurrent agents (default: 4)
-m, --model TEXT Decomposition model (default: opus)
-b, --budget FLOAT Max budget in USD (default: 5.0)
-r, --retry INTEGER Max retries for failed tasks (default: 1)
-c, --config PATH Path to swarm.yaml
--demo Run demo simulation (no API key needed)
--dry-run Show plan without executing
--quality-gate/--no-quality-gate Enable/disable Opus quality review (default: on)
--no-ui Disable rich terminal UI
-v, --version Show version
# Subcommands
claude-swarm sessions # List past sessions
claude-swarm replay <session-id> # Replay a session's events
YAML Configuration
Create swarm.yaml in your project root to define custom agent types:
swarm:
name: full-stack-review
max_concurrent: 4
budget_usd: 5.0
model: opus
agents:
security-reviewer:
description: Reviews code for OWASP vulnerabilities
model: opus
tools: [Read, Grep, Glob]
prompt: |
Analyze the code for SQL injection, XSS, CSRF...
tester:
description: Writes and runs tests
model: haiku
tools: [Read, Write, Edit, Bash]
prompt: |
Write comprehensive tests. Ensure 80% coverage...
connections:
- from: coder
to: security-reviewer
- from: coder
to: tester
- from: [security-reviewer, tester]
to: reviewer
Claude Swarm auto-detects swarm.yaml or .claude/swarm.yaml in your project.
How Opus 4.6 Is Used
Claude Swarm demonstrates strategic model selection with two critical Opus 4.6 touchpoints:
Phase 1: Task Decomposition (Planning)
Opus 4.6 handles the hardest reasoning task — analyzing your codebase, understanding the architecture, identifying dependencies between subtasks, and producing a parallelizable execution plan. This requires deep understanding of code relationships and optimal task splitting.
Phase 2: Worker Execution
Haiku handles the parallelizable work — each agent follows focused instructions from the plan. Using Haiku here is 3x cheaper while maintaining 90% of Sonnet's capability for focused tasks.
Phase 2.5: Quality Gate (Review)
After all agents complete, Opus 4.6 reviews the combined output. It checks for integration issues between agents' work, missed edge cases, security concerns, and whether the original task was fully addressed. This catches problems that individual agents can't see.
This mirrors real engineering team structure: a senior architect designs the plan, junior engineers execute in parallel, and the senior reviews the combined result.
Tech Stack
Python 3.11+ with anyio for structured async concurrency
claude-agent-sdk (v0.1.35+) for Claude Code subprocess control
Rich for terminal UI (Live dashboard with panels and tables)
Click for CLI framework
Pydantic for data validation
NetworkX for dependency graph topological sorting
Development
# Clone and install
git clone https://github.com/affaan-m/claude-swarm
cd claude-swarm
pip install -e ".[dev]"
# Run tests (44 passing)
pytest tests/ -v
# Lint
ruff check src/ tests/
Project Structure
src/claude_swarm/
cli.py CLI entry point (Click group + subcommands)
types.py Core dataclasses (SwarmTask, SwarmPlan, etc.)
decomposer.py Opus 4.6 task decomposition
orchestrator.py Parallel execution with file locks, budget, retries
quality_gate.py Opus 4.6 quality review of agent outputs
demo.py Demo simulation with animated TUI
config.py YAML swarm topology configuration
session.py JSONL event recording and replay
ui.py Rich terminal dashboard
License
MIT — Affaan Mustafa
Acknowledgments
Claude Agent SDK for the subprocess control layer
Everything Claude Code for agent patterns and inspiration
Built for the Cerebral Valley x Anthropic Claude Code Hackathon
Does affaan-m/claude-swarm have a project website?
No homepage URL was recorded for affaan-m/claude-swarm in TopGit's last sync. The README tab above frequently contains screenshots and demo links, or check the repository description on GitHub.
How active is development on affaan-m/claude-swarm?
The most recent commit recorded on affaan-m/claude-swarm was 8 months ago, based on the GitHub push timestamp. The repository has 103 forks — one of the better signals of community interest.
How many stars does affaan-m/claude-swarm have?
affaan-m/claude-swarm has 406 GitHub stars — refresh the page for the live number, or check github.com/affaan-m/claude-swarm. TopGit mirrors GitHub's count but does not claim minute-by-minute accuracy.
What language is affaan-m/claude-swarm written in?
affaan-m/claude-swarm is written primarily in Python. GitHub's language field is based on the largest share of bytes in the default branch.
What license does affaan-m/claude-swarm use?
affaan-m/claude-swarm is released under the MIT license. Always verify the LICENSE file directly on GitHub for the authoritative terms — license strings can be edited out of sync with a project's actual stance.
What topics is affaan-m/claude-swarm associated with?
GitHub's repository topics for affaan-m/claude-swarm: "agent-sdk", "agents", "ai", "claude", "claude-code", "hackathon", "multi-agent", "orchestration", "swarm". TopGit's editorial category is AI Tools.
Where do I read more about affaan-m/claude-swarm?
This TopGit page is a snapshot — the READ ME tab shows the project's own README content (links stripped, images preserved). The GitHub repository at github.com/affaan-m/claude-swarm is the definitive source.
Read full README in the tab above.
Want a second opinion on claude-swarm?
Ask an AI that can read this page — one click and you get its take on claude-swarm.
/claude-swarm
