How hfdaraijan/ruflo works
Ruflo is an agent harness for Claude Code and Codex: it routes work, coordinates agents, supplies memory and tools, and applies security controls. The principal documented flow runs from a CLI/MCP or editor interaction through routing and swarm coordination to agent work, with learning and memory supporting later tasks. The graph combines that documented flow with source-backed workflow, embedding, policy, federation, and UI components. Several links reflect README-level architecture because their implementation was not sampled.
Main entry points: plugins Specialist agents · generate.ts UI API · coverage-router.ts Task router · ruflo.js Ruflo CLI
Components 19
Entry points
- Ruflo CLI
ruflo/bin/ruflo.js - MCP bridge
ruflo/src/mcp-bridge/index.js - Hooks
plugin/scripts/ruflo-hook.cjs
Coordination
- Task router
v3/@claude-flow/cli/src/ruvector/coverage-router.ts - Workflow engine
v3/src/task-execution/application/WorkflowEngine.ts - Swarm coordination
plugins - Specialist agents
plugins
Agent capabilities
- Learning loop
plugins/ruflo-metaharness/scripts - Embedding service
v3/@claude-flow/embeddings/src/embedding-service.ts - Agent memory
plugins - LLM providers
Safety and federation
- Policy engine
v3/@claude-flow/security/src/policy/engine.ts - Federation policy
v3/@claude-flow/plugin-agent-federation/src/application/policy-engine.ts - PII pipeline
v3/@claude-flow/plugin-agent-federation/src/domain/services/pii-pipeline-service.ts
User surfaces
- Ruflo UI
ruflo/src/ruvocal/src/routes/+page.svelte - UI API
ruflo/src/ruvocal/src/lib/server/textGeneration/generate.ts - Conversation store
ruflo/src/ruvocal/src/lib/server/database.ts
Other components
- Engineer
- Claude Code / Codex
Connections 19
- Engineer → Ruflo CLI: invokes
- Claude Code / Codex → Hooks: triggers
- Claude Code / Codex → MCP bridge: calls tools
- Ruflo CLI → Task router: routes work
- Hooks → Task router: routes work
- MCP bridge → Swarm coordination: dispatches
- Task router → Swarm coordination: dispatches
- Swarm coordination → Specialist agents: coordinates
- Workflow engine → Specialist agents: executes tasks
- Specialist agents → Learning loop: feeds outcomes
- Learning loop → Agent memory: updates patterns
- Specialist agents → Agent memory: uses memory
- Specialist agents → LLM providers: requests inference
- Specialist agents → Policy engine: checks actions
- Federation policy → PII pipeline: applies filtering
- Engineer → Ruflo UI: uses
- Ruflo UI → UI API: sends requests
- UI API → Conversation store: reads/writes
- UI API → LLM providers: requests generation