swarm

# Swarm — Cut Your LLM Costs by 200x

**Turn your expensive model into an affordable daily driver. Offload the boring stuff to Gemini Flash workers — parallel, batch, research — at a fraction of the cost.**

## At a Glance

| 30 tasks via | Time | Cost |
|--------------|------|------|
| Opus (sequential) | ~30s | ~$0.50 |
| Swarm (parallel) | ~1s | ~$0.003 |

## When to Use

Swarm is ideal for:
- **3+ independent tasks** (research, summaries, comparisons)
- **Comparing or researching multiple subjects**
- **Multiple URLs** to fetch/analyze
- **Batch processing** (documents, entities, facts)
- **Complex analysis** needing multiple perspectives → use chain

## Quick Reference

```bash
# Check daemon (do this every session)
swarm status

# Start if not running
swarm start

# Parallel prompts
swarm parallel "What is X?" "What is Y?" "What is Z?"

# Research multiple subjects
swarm research "OpenAI" "Anthropic" "Mistral" --topic "AI safety"

# Discover capabilities
swarm capabilities
```

## Execution Modes

### Parallel (v1.0)
N prompts → N workers simultaneously. Best for independent tasks.

```bash
swarm parallel "prompt1" "prompt2" "prompt3"
```

### Research (v1.1)
Multi-phase: search → fetch → analyze. Uses Google Search grounding.

```bash
swarm research "Buildertrend" "Jobber" --topic "pricing 2026"
```

### Chain (v1.3) — Refinement Pipelines
Data flows through multiple stages, each with a different perspective/filter. Stages run in sequence; tasks within a stage run in parallel.

**Stage modes:**
- `parallel` — N inputs → N workers (same perspective)
- `single` — merged input → 1 worker
- `fan-out` — 1 input → N workers with DIFFERENT perspectives
- `reduce` — N inputs → 1 synthesized output

**Auto-chain** — describe what you want, get an optimal pipeline:
```bash
curl -X POST http://localhost:9999/chain/auto \
  -d '{"task":"Find business opportunities","data":"...market data...","depth":"standard"}'
```

**Manual chain:**
```bash
swarm chain pipeline.json
#...

# 🐝 Swarm — Parallel Task Execution for OpenClaw

[![Version](https://img.shields.io/badge/version-1.2.0-blue.svg)](./CHANGELOG.md)

**Turn sequential LLM tasks into parallel operations. 200x cheaper than Opus. Now with live web search.**

## What is Swarm?

Swarm adds parallel processing to [OpenClaw](https://github.com/openclaw/openclaw) by distributing work across cheap Gemini Flash workers. Instead of burning expensive tokens sequentially, fire 30 tasks in parallel for $0.003.

```
┌────────────────────────────────────────────────────┐
│              Coordinator (Claude)                   │
│         Orchestration • Memory • Synthesis          │
└─────────────────────┬──────────────────────────────┘
                      │ Task Distribution
    ┌─────────────────┼─────────────────┐
    │                 │                 │
┌───▼───┐        ┌───▼───┐        ┌───▼───┐
│Search │        │ Fetch │        │Analyze│
│Workers│        │Workers│        │Workers│
│(Flash)│        │(Flash)│        │(Flash)│
│  🔍   │        │  📄   │        │  🧠   │
└───────┘        └───────┘        └───────┘
      Optional: Google Search grounding
```

### What's New in v1.2.0

- 💰 **Cost tracking & savings reports** — See exactly how much you're saving vs Opus in real-time
- 🔄 **Auto-retry** — Failed tasks retry automatically with backoff (skips non-transient errors)
- ⏱️ **Task timeouts** — 30s default prevents hung workers from blocking the pool
- 🧹 **Dead code cleanup** — Removed 1,300+ lines of unused code
- 🔍 **Web Search Grounding** — Workers search the live web via Google Search (Gemini only, no extra cost)
- 🔒 **Security policy** — Workers answer research questions while blocking credential exfiltration

### Performance

| Tasks | Sequential (Opus) | Swarm (Parallel) | Cost Savings |
|-------|-------------------|------------------|--------------|
| 5 | ~5s, $0.08 | **1.5s, $0.0005** | 160x cheaper |
| 10 | ~10s, $0.17 | **1.5s, $0.001** | 170x cheaper |
| 30 | ~30s, $0.50 |...