Skip to content

armgit / ruflo

Public Imported from GitHub
0

🌊 The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated

Checkpoint: File edits

rUve723686edf9a
diff --git a/v3/@claude-flow/cli/.claude/commands/swarm/swarm.md b/v3/@claude-flow/cli/.claude/commands/swarm/swarm.md
index 48b69378cf..6912c41629 100644
--- a/v3/@claude-flow/cli/.claude/commands/swarm/swarm.md
+++ b/v3/@claude-flow/cli/.claude/commands/swarm/swarm.md
@@ -1,27 +1,58 @@
# swarm
-Main swarm orchestration command for Claude Flow.
+Main swarm orchestration command for Claude Flow V3.
+
+## 🚨 CRITICAL: Background Execution Pattern
+
+**When spawning a swarm, Claude Code MUST:**
+
+1. **Spawn ALL agents in background** using `run_in_background: true`
+2. **Put ALL Task calls in ONE message** for parallel execution
+3. **Tell user what's happening** with clear agent list
+4. **STOP and WAIT** - don't add more tool calls or poll status
+
+```javascript
+// ✅ CORRECT: Background parallel agents in ONE message
+Task({ prompt: "Research...", subagent_type: "researcher", run_in_background: true })
+Task({ prompt: "Code...", subagent_type: "coder", run_in_background: true })
+Task({ prompt: "Test...", subagent_type: "tester", run_in_background: true })
+
+// Then TELL USER and STOP:
+"I've launched 3 agents working in parallel. They'll report when done."
+```
+
+```javascript
+// ❌ WRONG: Don't poll or check status
+TaskOutput({ task_id: "..." }) // NO!
+swarm status // NO!
+"Should I check on the agents?" // NO!
+```
## Usage
```bash
-npx claude-flow swarm <objective> [options]
+npx @claude-flow/cli@latest swarm <objective> [options]
```
## Options
- `--strategy <type>` - Execution strategy (research, development, analysis, testing)
-- `--mode <type>` - Coordination mode (centralized, distributed, hierarchical, mesh)
-- `--max-agents <n>` - Maximum number of agents (default: 5)
-- `--claude` - Open Claude Code CLI with swarm prompt
-- `--parallel` - Enable parallel execution
+- `--topology <type>` - Swarm topology (hierarchical, mesh, ring, star)
+- `--max-agents <n>` - Maximum number of agents (default: 15)
+- `--background` - Run agents in background (default: true)
+- `--parallel` - Enable parallel execution (default: true)
## Examples
```bash
-# Basic swarm
-npx claude-flow swarm "Build REST API"
+# Initialize swarm
+npx @claude-flow/cli@latest swarm init --topology hierarchical
-# With strategy
-npx claude-flow swarm "Research AI patterns" --strategy research
-
-# Open in Claude Code
-npx claude-flow swarm "Build API" --claude
+# Check status (ONLY when agents complete)
+npx @claude-flow/cli@latest swarm status
```
+
+## Agent Types by Task
+| Task | Agents |
+|------|--------|
+| New Feature | researcher, architect, coder, tester, reviewer |
+| Bug Fix | researcher, coder, tester |
+| Refactor | architect, coder, reviewer |
+| Security | security-architect, security-auditor, reviewer |