Effin.AI

The Most Powerful AI Agent Orchestration CLI Ever Built.

586 MCP tools. 223 agent types. 98 skills. Self-learning swarms. Anti-drift consensus. Autonomous PRD-to-code loops. Meta-orchestration. Stealth browser automation. Intelligent web scraping. 500+ app integrations. Domain-specific AI verticals. All in one install. Two dependencies.

One-Line Install

curl -fsSL https://raw.githubusercontent.com/effinai/effinai/main/scripts/install.sh | bash

Or via npm/npx:

npm install -g effinai        # Global install
npx effinai@latest init       # Or run directly

What Makes Effin.AI Different

| Feature | Other Tools | Effin.AI | |---------|-----------|---------| | MCP Tools | 5-20 tools | 570 native MCP tools — the largest MCP toolset ever shipped | | Agent Types | 5-10 generic agents | 223 specialized agents — engineering, design, sales, legal, healthcare, game dev, spatial computing, finance, marketing, and more | | Self-Learning | Static, no adaptation | Pattern memory that makes agents smarter with every task (boost/decay scoring) | | Anti-Drift | None — agents wander | Real-time drift detection + automatic realignment (40% threshold) | | Meta-Orchestration | Single orchestrator | Orchestrate multiple orchestrators — coordinate entire swarm fleets | | Autonomous Mode | Manual orchestration | effin run --prd tasks.json — PRD to production code, fully autonomous | | Auto-Orchestration | Manual setup | Describe what you want — Effin.AI auto-selects agents, skills, topology, and model tier | | Skill Curator | Static skill libraries | Auto-creates skills from successful tasks, archives stale ones | | Model Routing | One model fits all | 3-tier routing: Booster (<1ms, $0) / Fast (~500ms) / Full (2-5s) | | App Integrations | Manual API wiring | 500+ apps via native integration (Slack, GitHub, Jira, Notion, and more) | | Domain Verticals | Generic AI | Financial Risk, Healthcare Clinical, Legal Contracts, IoT, Quantum Optimization, Neural Trading | | Agent Swarms | Single agent | 5 topologies with consensus (raft, quorum, gossip) + file-based messaging | | Memory | Session-only | Persistent vector memory with HNSW indexing + TF-IDF embeddings + pattern learning | | Web Scraping | Basic fetch | Anti-bot bypass with stealth fingerprinting | | Browser | Playwright basic | Stealth browser — undetectable automation | | Dependencies | 50+ packages | 2 runtime deps (semver, zod). Installs in 10 seconds. | | Magic Router | Command memorization | Natural language — just type what you want |

Quick Start

# Initialize in your project
effin init

# See the rich status dashboard
effin status

# Spawn an agent
effin agent spawn --type coder --name my-coder --task "Add login endpoint"

# Start a multi-agent swarm
effin swarm init --topology hierarchical --max-agents 8

# Autonomous PRD-to-code (like having a dev team)
effin run --prd tasks.json

# Search learned patterns
effin memory search --query "authentication patterns"

# Auto-orchestrate (no manual setup needed)
# Just describe the task — Effin.AI does the rest
effin "build a payment system with Stripe integration"

# Meta-orchestrate multiple swarms
effin multica orchestrate --task "Build entire microservices platform"

# Multi-agent debate simulation
effin simulate --scenario "Should we migrate to microservices?"

# AI-powered trend prediction
effin predict --topic "AI agent adoption" --horizon long

# Check system health
effin doctor

Architecture

┌──────────────────────────────────────────────────────────────────────┐
│                           Effin.AI CLI                               │
│  19 commands · 50+ subcommands · 23+ magic routes · MCP auto-detect  │
├──────────────────────────────────────────────────────────────────────┤
│  META-ORCHESTRATION                                                  │
│  Orchestrate multiple Effin.AI instances across organizations        │
├──────────────────────────────────────────────────────────────────────┤
│  AUTO-ORCHESTRATOR                                                   │
│  Analyzes task → selects topology, agents, skills, model tier        │
├──────────────────────────────────────────────────────────────────────┤
│  SWARM COORDINATOR                                                   │
│  5 topologies · raft/quorum/gossip consensus · anti-drift detection  │
├──────────┬──────────┬──────────┬──────────┬──────────┬──────────────┤
│ AGENTS   │ SKILLS   │ MEMORY   │ LEARNING │ AUTONOMY │ SIMULATION   │
│ 223 spec-│ 98 built │ JSON +   │ Pattern  │ PRD loop │ Multi-agent  │
│ ialized  │ -in with │ HNSW     │ extract  │ Quality  │ debate &     │
│ types    │ progres- │ vector   │ boost/   │ gates    │ prediction   │
│          │ sive load│ search   │ decay    │ Git auto │              │
├──────────┴──────────┴──────────┴──────────┴──────────┴──────────────┤
│  MCP SERVER (stdio) — 570 tools for Claude Code                      │
├──────────────────────────────────────────────────────────────────────┤
│  INTEGRATIONS: 500+ apps · 8 model providers · stealth browser/scrape│
├──────────────────────────────────────────────────────────────────────┤
│  DOMAIN VERTICALS: Finance · Healthcare · Legal · IoT · Quantum      │
└──────────────────────────────────────────────────────────────────────┘

CLI Commands (19 Commands)

| Command | Subcommands | What It Does | |---------|-------------|-------------| | effin init | — | Initialize project, scaffold .effinai/ directory | | effin agent | spawn\|list\|stop\|types\|logs | Agent lifecycle — spawn any of 223 types | | effin swarm | init\|status\|stop\|route\|presets | Multi-agent swarm coordination | | effin skill | list\|run\|create\|info | Progressive skill management (98 skills) | | effin memory | store\|search\|list\|retrieve\|stats | Persistent vector memory with HNSW | | effin run | --prd file.json | Autonomous PRD-to-code execution loop | | effin hooks | pre-task\|post-task\|route\|status | Self-learning lifecycle hooks | | effin curator | status\|run\|pin | Skill lifecycle management | | effin config | get\|set\|list | Configuration management | | effin status | — | Rich multi-line status dashboard | | effin doctor | --fix | Health checks and auto-repair | | effin mcp | start\|info | MCP server for Claude Code | | effin tools | list\|scrape | External tools (stealth scraper, browser) | | effin providers | list | Multi-model provider management | | effin simulate | --scenario | Multi-persona swarm debate simulation | | effin predict | --topic --horizon | AI-powered trend prediction and forecasting | | effin workspace | create\|list\|merge | Git worktree isolation per agent | | effin multica | connect\|status\|agents\|orchestrate\|... | Meta-orchestration of multiple swarms | | effin /<command> | 16 slash commands | /prd /tdd /review /security /deploy and more |

223 Built-In Agent Types

Effin.AI ships with the largest library of specialized AI agents ever assembled. Every agent has a unique persona, system prompt, and model tier recommendation.

Core Development (20+)

coder senior-developer backend-dev frontend-dev mobile-dev debugger refactorer rapid-prototyper minimal-change-engineer code-analyzer codebase-onboarding-engineer embedded-engineer embedded-firmware-engineer kernel-developer compiler-engineer solidity-engineer blockchain-dev

Architecture & Design (10+)

architect system-architect cloud-architect api-designer database-admin dba-postgres dba-mysql design-system ux-designer ui-designer color-specialist typography-expert motion-designer

Testing & Quality (8+)

tester qa-engineer test-automation reviewer penetration-tester performance-benchmarker performance-engineer reliability-engineer

DevOps & Infrastructure (12+)

devops-engineer cicd-engineer kubernetes-specialist terraform-specialist monitoring-specialist incident-responder cache-specialist redis-specialist kafka-specialist network-engineer terminal-integration-specialist

Security (6+)

security-auditor security-architect threat-modeler threat-detection-engineer compliance-auditor privacy-officer

AI & Machine Learning (8+)

ml-developer mlops-engineer ai-prompt-engineer ai-safety-researcher computer-vision nlp-specialist llm-fine-tuner voice-ai-engineer rag-architect

Data & Analytics (8+)

data-engineer data-analyst data-scientist bi-developer etl-developer analytics-specialist search-engineer search-query-analyst

API & Integration (8+)

api-docs api-tester graphql-specialist webhook-specialist integration-specialist realtime-engineer lsp-index-engineer email-intelligence-engineer

Documentation & Writing (5+)

documentation technical-writer medical-writer developer-advocate technical-evangelist

Project Management (6+)

project-manager product-manager scrum-master stakeholder-manager operations-manager meeting-facilitator

Game Development (10+)

game-designer game-artist game-audio game-audio-engineer level-designer narrative-designer multiplayer-engineer godot-developer godot-gameplay-scripter technical-artist unity-architect unity-developer unreal-developer unreal-systems-engineer roblox-experience-designer roblox-systems-scripter roblox-avatar-creator

Spatial Computing & XR (6+)

xr-developer xr-interface-architect xr-cockpit-interaction-specialist webxr-developer visionos-engineer visionos-spatial-engineer metaverse-architect spatial-ui-designer macos-metal-engineer

Sales & Business Development (12+)

sales-strategist sales-engineer sales-deal-strategist sales-discovery-coach sales-outbound-strategist sales-pipeline-analyst sales-proposal-strategist deal-closer discovery-coach outbound-strategist proposal-writer account-manager success-manager

Marketing & Growth (12+)

growth-hacker growth-marketer brand-strategist brand-guardian content-strategist copywriter seo-specialist social-media-manager social-media-strategist email-marketer email-marketing-specialist demand-gen paid-media-auditor paid-social-strategist ppc-strategist podcast-strategist programmatic-buyer ad-creative-strategist community-manager visual-storyteller video-producer

Finance & Legal (12+)

financial-analyst fpa-analyst budget-planner tax-strategist bookkeeper bookkeeper-controller revenue-ops cost-optimizer investment-researcher legal-analyst legal-compliance-specialist legal-document-reviewer contract-specialist ip-specialist regulatory-analyst

Healthcare & Science (6+)

clinical-analyst health-informatics healthcare-support-specialist medical-writer anthropologist psychologist narratologist historian geographer

HR & Operations (8+)

recruiter recruitment-specialist hr-specialist hr-onboarding-specialist onboarding-specialist training-specialist culture-architect chief-of-staff customer-service-agent escalation-handler support-agent facilities-planner supply-chain-analyst supply-chain-strategist procurement-specialist hospitality-guest-services

Swarm & Orchestration (5+)

coordinator sparc-coordinator consensus-builder swarm-intelligence optimizer agent-builder workflow-designer tool-evaluator reality-checker

Specialized (10+)

i18n-specialist accessibility migration-specialist multi-repo-specialist dependency-manager git-specialist pr-manager release-manager repo-architect issue-tracker tracking-specialist knowledge-base-curator identity-graph-operator image-prompt-engineer filament-optimization-specialist real-estate-analyst automation-governance-architect cultural-intelligence-strategist

98 Built-In Skills

Skills are reusable workflows that agents can execute. Each skill has progressive loading — only frontmatter (~100 tokens) is loaded for listing; full body loads on demand.

Development & Code

code-review tdd lint-fix refactor-safe type-safety quick-fix feature-flag batch-executor background-tasks exa-codesearch agent-grep agentgrep code-search codebase-onboarding

Architecture & Design

architecture-review adr-lifecycle adr-writer ddd-design spec-driven sparc-methodology database-designer database-schema

Testing & Quality

test-generator test-coverage webapp-testing performance-audit performance-profiler security-scan threat-hunting pii-detection dependency-audit

AI & Agents

agent-designer self-improving-agent selfdev-harness self-dev ambient-agent ambient-memory agenthub skills-hub mcp-builder karpathy-coder rag-architect llm-wiki trajectory-learning

Memory & Knowledge

knowledge-graph graph-rag hybrid-search memory-graph holographic-memory context-compaction

DevOps & Infrastructure

ci-pipeline docker-compose k8s-deploy monitoring-setup incident-response release-workflow git-workflow migration-plan migration-planner

Documentation & Content

changelog readme-generator api-docs-generator onboarding-guide content-research-writer developer-growth

Productivity & Workflow

prd-generator autopilot-loop goal-loop goal-tracker cron-scheduler session-resume session-search shell-hooks file-organizer n8n-workflow

Integration & Communication

composio-connect multi-platform-gateway webhook-gateway federation-protocol browser-agent lead-research meeting-insights

Business & Analytics

cost-analysis cost-optimizer cost-tracking invoice-organizer observability sidecar-verifier

Document Processing

pdf-toolkit docx-processor xlsx-processor

Research & Exploration

autoresearch swarm-coordinator cache-optimizer api-generator

570 MCP Tools (Claude Code Integration)

Effin.AI ships the largest MCP toolset ever built. Register once, get 570 tools in Claude Code:

claude mcp add effin -- npx -y effinai@latest

Tool Categories

| Category | Count | Examples | |----------|-------|---------| | Core CLI | 10 | swarm_init, agent_spawn, agent_list, status, route, doctor | | Agent Management | 7 | agent_types, agent_logs, agent_assign, agent_health, agent_pool, broadcast | | Memory & Learning | 14 | memory_store, memory_search, pattern_store, pattern_boost, learning_stats | | Skills | 5 | skill_list, skill_run, skill_create, skill_search, skill_info | | Swarm Coordination | 5 | swarm_topology, swarm_agents, swarm_consensus, team_create | | External Tools | 2 | tools_list, scrape | | Simulation & Analysis | 4 | simulate, predict, brainstorm, compare | | Workspace & Config | 6 | workspace_create, workspace_merge, config_get, providers_list | | Hooks & Lifecycle | 3 | hooks_fire, hooks_list, curator_status | | Security & Compliance | 5 | security_scan, pii_detect, compliance_check, threat_model, audit_log | | Cost & Performance | 4 | cost_track, cost_report, budget_check, benchmark | | Architecture (SPARC/ADR/DDD) | 5 | sparc_phase, adr_create, ddd_context, knowledge_graph | | Autonomous Loops | 4 | goal_plan, loop_start, loop_status, federation_register | | Observability | 3 | trace_start, metrics_report, audit_trail | | Auto-Orchestration | 1 | orchestrate (intelligent auto-selection of everything) | | App Integration | 5 | composio_apps, composio_actions, composio_execute, composio_connect | | AgentDB | 11 | agentdb_store, agentdb_query, agentdb_vector_search, cache ops | | AI Defense | 3 | aidefence_scan, validate, report | | Autopilot | 9 | autopilot_start, plan, execute, evaluate, iterate, stop, status | | Browser Sessions | 3 | browser_open, navigate, extract | | Diff & Patch | 3 | diff_generate, apply, review | | Federation | 13 | federation_init, peer_connect, sync, broadcast, consensus ops | | Intelligence & Neural | 8 | intelligence_analyze, pattern_detect, suggest, neural_train | | Knowledge Graph | 5 | kg_create, query, traverse, visualize, export | | Worker & Scheduling | 7 | worker_spawn, assign, schedule, pool management | | Market Data | 5 | market_analyze, portfolio, risk assessment | | Migration | 6 | migration_plan, execute, rollback, validate, status | | Neural Trader | 9 | neural_trade, backtest, signals, portfolio, risk | | Observability Extended | 5 | trace, metrics, logs, dashboard, alerts | | Plugin Creator | 2 | plugin_create, publish | | RAG Memory | 6 | rag_ingest, search, chunk, embed, rerank, summarize | | Vector Intelligence | 12 | vector_embed, search, cluster, reduce, visualize | | WASM Execution | 3 | wasm_load, execute, debug | | Workflow Engine | 4 | workflow_create, execute, status, template | | ADR Management | 5 | adr_propose, review, accept, supersede, search | | SONA Intelligence | 14 | sona_adapt, trajectory learning, pattern ops, optimize | | Task Management | 8 | task_create, assign, prioritize, dependencies, track | | Teammate Plugin | 21 | teammate_spawn, assign, review, collaborate, handoff | | Project Orchestration | 17 | project_init, plan, milestone, dependency, resource ops | | Mathematical AI | 6 | math_solve, prove, conjecture, optimize, visualize | | Code Intelligence v3 | 5 | code_analyze, suggest, refactor, explain, dependencies | | Cognitive Kernel | 5 | cognitive_reason, plan, reflect, adapt, learn | | IoT Fleet | 22 | iot_device, fleet, telemetry, command, firmware, diagnostics | | Financial Risk | 5 | risk_assess, var_calculate, stress_test, scenario, report | | Healthcare Clinical | 5 | clinical_analyze, protocol, risk, pathway, report | | Hyperbolic Reasoning | 5 | reason_embed, navigate, cluster, hierarchy, visualize | | Legal Contracts | 5 | contract_analyze, clause, risk, compare, generate | | Performance Optimizer | 5 | perf_profile, bottleneck, optimize, benchmark, report | | Quantum Optimizer | 5 | quantum_optimize, anneal, variational, circuit, benchmark | | Test Intelligence | 5 | test_generate, mutate, coverage, prioritize, report | | Coherence Gate | 3 | coherence_check, align, enforce | | PRD Execution | 3 | prd_parse, execute, validate | | Engineering Tools | 7 | ci_analyze, dependency_audit, schema_migrate, feature_flag | | Wiki & Knowledge | 5 | wiki_create, search, update, link, export | | Product & Business | 9 | product_roadmap, user_research, competitive_analysis, pricing | | Accessibility | 2 | a11y_audit, remediate | | Self-Improvement | 1 | self_improve | | Codebase Intelligence | 16 | nexus_analyze, nexus_query, nexus_search, nexus_impact, nexus_self_analyze | | Meta-Orchestration | 12 | multica_status, connect, agents, runtimes, orchestrate, autopilots |

Codebase Intelligence (Self-Aware AI)

Effin.AI can understand its own codebase — and yours. Built-in knowledge graph indexing turns any repository into a queryable graph of symbols, dependencies, call chains, and clusters.

# Index your codebase into a knowledge graph
effin tools install gitnexus
npx gitnexus analyze

# Query via MCP tools (available in Claude Code)
# effinai_nexus_query "find all functions that call the auth middleware"
# effinai_nexus_impact "src/core/swarm-coordinator.ts"
# effinai_nexus_clusters --format mermaid

# Self-analysis — Effin.AI understands its own architecture
# effinai_nexus_self_analyze --aspect full
# effinai_nexus_explain --capability "anti-drift"

# Visual exploration via web UI
npx gitnexus serve    # Opens browser-based graph explorer

What Codebase Intelligence Provides

• Knowledge Graph — Every symbol, dependency, and call chain indexed
• Hybrid Search — BM25 keyword + semantic embeddings + reciprocal rank fusion
• Impact Analysis — Know exactly what breaks before you change anything
• Cluster Detection — Leiden algorithm identifies natural module boundaries
• Architecture Docs — Auto-generated Mermaid diagrams from the graph
• Self-Awareness — Effin.AI can explain its own capabilities and code paths
• 14 Languages — TypeScript, JavaScript, Python, Java, Go, Rust, C/C++, and more
• Web UI — Browser-based WebGL graph visualization (Sigma.js)

Self-Analysis MCP Tools (16 tools)

| Tool | What It Does | |------|-------------| | nexus_analyze | Index codebase into knowledge graph | | nexus_query | Natural language or Cypher graph queries | | nexus_search | Hybrid search across indexed code | | nexus_context | Rich context for any symbol (definition, usages, deps) | | nexus_impact | Impact analysis — what would be affected by a change | | nexus_dependencies | Full dependency graph for file/module | | nexus_clusters | Detected code clusters (Leiden community detection) | | nexus_callgraph | Call graph for any function | | nexus_rename | Coordinated rename with all references | | nexus_changes | Detect changes since last index | | nexus_architecture | Auto-generate architecture docs with Mermaid | | nexus_wiki | LLM-powered documentation generation | | nexus_stats | Knowledge graph statistics | | nexus_serve | Start web UI for visual graph exploration | | nexus_self_analyze | Effin.AI analyzes its own architecture and capabilities | | nexus_explain | Explain how a specific Effin.AI capability works |

Meta-Orchestration

Orchestrate multiple Effin.AI instances — the only system that lets you coordinate orchestrators:

# Connect to orchestration platform
effin multica connect --url https://your-instance.example.com --token sk-...

# View all connected agents and runtimes
effin multica status
effin multica agents
effin multica runtimes

# Meta-orchestrate: break complex task across multiple swarm fleets
effin multica orchestrate --task "Build entire microservices platform"

# Check multi-swarm orchestration progress
effin multica orchestration-status --id <orchestration-id>

# Manage autopilot workflows
effin multica autopilots

The meta-orchestrator automatically:

1. Breaks complex tasks into subtasks (Research, Architect, Implement, Test, Review)
2. Distributes subtasks across multiple registered Effin.AI runtimes
3. Assigns to optimal agents based on capability matching
4. Tracks cross-swarm progress with unified status

Auto-Orchestration

No manual setup needed. Just describe what you want:

effin "build a payment system with Stripe"

The auto-orchestrator:

1. Analyzes the task description for complexity, domain, and requirements
2. Selects topology — hierarchical for most tasks, mesh for research, adaptive for unknowns
3. Picks agents — the optimal combination from 223 types
4. Chooses skills — relevant skills from the 98 built-in library
5. Routes model — assigns the right model tier (Booster/Fast/Full) per agent
6. Executes with anti-drift monitoring and consensus

Self-Learning System

Effin.AI gets smarter every time you use it:

New Task --> Search Patterns --> Inject Context --> Execute --> Quality Gate
    ^                                                             |
    +-- Store Pattern (if success, boost +20%) <------------------+
    +-- Decay Pattern (if stale, -10%) <-- Curator lifecycle review

1. RETRIEVE — HNSW vector search finds similar past successes via TF-IDF embeddings
2. EXECUTE — Agents work with retrieved context automatically injected
3. JUDGE — Quality gate (typecheck/lint/test) provides success/failure signal
4. STORE — Successful patterns saved with embeddings and boost score
5. DECAY — Stale patterns automatically decay; unused skills get archived

After 3+ successful patterns, the system provides relevant context to agents automatically. Pattern scores range from 0-100 with +20 boost on success, -10 decay on staleness.

Swarm Topologies

effin swarm init --topology hierarchical --max-agents 8 --strategy specialized

| Topology | Best For | How It Works | |----------|---------|-------------| | hierarchical | Most tasks (anti-drift) | Coordinator dispatches, workers report back | | mesh | Research/exploration | All agents communicate freely via file-based messaging | | ring | Pipeline tasks (A->B->C) | Sequential pass-through | | star | Fan-out/fan-in | Central hub with spokes | | adaptive | Unknown complexity | Starts hierarchical, adapts topology as needed |

Consensus Mechanisms

• Raft — Leader-based, tolerates f < n/2 failures (default)
• Quorum — Configurable quorum threshold for decisions
• Gossip — Epidemic protocol for eventual consistency in large swarms

Anti-Drift Detection

The only multi-agent system that actively monitors and corrects agent drift:

1. Each task has a task hash from the original description
2. Agent outputs are compared against task keywords (overlap scoring)
3. If divergence exceeds 40%, coordinator sends realignment message
4. After 2 failed realignments, task is reassigned to a new agent
5. All drift events are logged for system improvement

Agent Presets (Auto-Routing)

effin swarm route --task "fix the login bug"      # -> bugfix preset
effin swarm route --task "add payment feature"     # -> feature preset
effin swarm route --task "audit for XSS"           # -> security preset
effin swarm route --task "optimize database"       # -> performance preset

| Preset | Agents Spawned | |--------|---------------| | bugfix | researcher, coder, tester | | feature | architect, coder, tester, reviewer | | refactor | architect, coder, reviewer | | performance | optimizer, coder, tester | | security | security-auditor, coder, reviewer | | docs | researcher, coder | | full | coordinator, researcher, architect, coder, tester, reviewer |

3-Tier Model Routing

Every task is automatically routed to the optimal cost/performance tier:

effin hooks pre-task --description "rename variable foo to bar"
# -> [AGENT_BOOSTER_AVAILABLE] Tier 1: Skip LLM, <1ms, $0

effin hooks pre-task --description "fix the null check bug"
# -> [TASK_MODEL_RECOMMENDATION] Tier 2: haiku, ~500ms, $0.0002

effin hooks pre-task --description "design the auth architecture"
# -> [TASK_MODEL_RECOMMENDATION] Tier 3: opus, 2-5s, $0.015

| Tier | Handler | Latency | Cost | Use Cases | |------|---------|---------|------|-----------| | 1 | Agent Booster | <1ms | $0 | Simple transforms (var->const, add-types, remove-console) | | 2 | Fast Model | ~500ms | $0.0002 | Bug fixes, simple tasks, low complexity | | 3 | Full Model | 2-5s | $0.003-$0.015 | Architecture, security, complex reasoning |

8 Model Providers

| Provider | Models | |----------|--------| | Anthropic | Claude Opus, Sonnet, Haiku | | OpenAI | GPT-4o, GPT-4, GPT-3.5 | | Google | Gemini Pro, Gemini Flash | | xAI | Grok | | Kimi | Moonshot | | DeepSeek | DeepSeek Chat, Coder | | Ollama | Any local model | | LM Studio | Any local model |

effin providers list    # See all configured providers

Autonomous Mode (PRD-to-Code)

# Generate example PRD
effin run --example

# Edit prd.json with your stories, then:
effin run --prd prd.json --max-iterations 20

The autonomous loop:

1. Reads prd.json with user stories and acceptance criteria
2. Picks highest-priority pending story
3. Searches memory for similar past patterns
4. Spawns the right agents (based on task type)
5. Implements the story with anti-drift monitoring
6. Runs quality gates (typecheck, lint, test)
7. Commits on success: feat: [S1] Add login endpoint
8. Stores the pattern for future learning
9. Moves to next story
10. Repeats until all stories pass or max iterations reached

Simulation & Prediction

Multi-Agent Debate Simulation

effin simulate --scenario "Should we migrate to microservices?"

Spawns multiple AI personas (Optimist, Skeptic, Pragmatist) to debate the scenario, reaching a synthesized conclusion with confidence scores.

AI-Powered Trend Prediction

effin predict --topic "AI agent adoption" --horizon long

Generates predictions across short/medium/long horizons with confidence levels and key factors analysis.

500+ App Integrations

Native integration with 500+ apps across 15 categories via built-in connector:

| Category | Example Apps | |----------|-------------| | Dev Tools | GitHub, GitLab, Bitbucket, Jira, Linear, Notion | | Communication | Slack, Discord, Teams, Email | | Cloud | AWS, GCP, Azure, Vercel, Netlify | | Data | PostgreSQL, MongoDB, Redis, Elasticsearch | | CRM | Salesforce, HubSpot, Pipedrive | | Productivity | Google Workspace, Airtable, Trello | | Finance | Stripe, QuickBooks, Xero | | Marketing | Mailchimp, SendGrid, Twilio | | Storage | S3, GCS, Dropbox, Google Drive | | Monitoring | Datadog, PagerDuty, Sentry | | Auth | Auth0, Okta, Firebase Auth | | AI/ML | OpenAI, Anthropic, Hugging Face | | Social | Twitter/X, LinkedIn, Facebook | | E-Commerce | Shopify, WooCommerce | | Analytics | Mixpanel, Amplitude, Segment |

effin tools list    # See all available integrations

Set COMPOSIO_API_KEY in your environment to enable external app execution.

Domain Verticals

Effin.AI includes specialized domain intelligence:

Financial Risk Analysis

• Value-at-Risk calculations, stress testing, scenario analysis
• Portfolio risk assessment with Monte Carlo simulation
• Regulatory compliance reporting

Healthcare Clinical

• Clinical protocol analysis and pathway optimization
• Risk stratification and patient outcome prediction
• HIPAA-aware data handling

Legal Contract Intelligence

• Contract clause analysis and risk scoring
• Comparison across document versions
• Auto-generation of standard clauses

IoT Fleet Management

• Device telemetry monitoring and anomaly detection
• Fleet-wide firmware management and OTA updates
• Edge computing orchestration with 22 specialized tools

Quantum Optimization

• Quantum annealing and variational algorithms
• Circuit optimization and benchmark comparison
• Hybrid quantum-classical workflow support

Neural Trading

• Market signal detection and backtesting
• Portfolio optimization with risk constraints
• Multi-timeframe analysis

Magic Router (Natural Language)

No need to memorize commands — just type naturally:

effin "spawn a coder agent"           # -> agent spawn --type coder
effin "what if we deploy on Friday?"  # -> simulate
effin "predict AI adoption trends"    # -> predict
effin "search memory for auth"        # -> memory search
effin "show me all agents"            # -> agent list
effin "build a login system"          # -> run --task
effin "remember that JWT works best"  # -> memory store
effin "check system health"           # -> doctor
effin "orchestrate orchestrators"     # -> multica

23+ intent patterns with automatic argument extraction. Confidence scoring ensures accurate routing.

16 Slash Commands

effin /prd            # Generate PRD from description
effin /tdd            # Test-driven development workflow
effin /review         # Code review current changes
effin /changelog      # Generate changelog from commits
effin /security       # Security audit
effin /performance    # Performance profiling
effin /refactor       # Safe refactoring with tests
effin /deploy         # Deployment workflow
effin /test           # Generate and run tests
effin /docs           # Auto-generate documentation
effin /fix            # Quick bug fix
effin /architect      # Architecture design
effin /status         # System status
effin /learn          # Store learned pattern
effin /simulate       # Multi-agent simulation
effin /predict        # Trend prediction

Skill Curator (Auto-Evolution)

effin curator status    # See all skill lifecycle states
effin curator run       # Run lifecycle review
effin curator pin auth  # Prevent a skill from being archived

• Active — Recently used skills
• Stale (30+ days unused) — Marked for review
• Archived (90+ days unused) — Moved out of active rotation
• Pinned — Protected from archival
• Auto-Created — System generates skills from successful complex tasks

Ultrathink Mode

All agents use extended thinking (ultrathink) by default for maximum quality reasoning. This enables deeper analysis, better code generation, and more thoughtful architecture decisions across all 223 agent types.

Web Tools (Built-in)

Stealth Web Scraping

• Bypasses Cloudflare, WAFs, and bot detection
• Stealth fingerprinting with realistic browser profiles
• Automatic retry with rotation
• CSS/XPath selectors with smart extraction

Stealth Browser Automation

• Undetectable browser automation
• Custom fingerprint spoofing
• Passes all bot detection tests
• Integrates with agent workflows for UI testing

effin tools scrape https://example.com    # Scrape with anti-bot bypass
effin tools list                          # See all available tools

Rich Status Dashboard

effin status

Shows a comprehensive multi-line dashboard with:

• Version, git branch, MCP status
• Swarm topology, agent count, health percentage, anti-drift status
• Memory entries, patterns, learning status, HNSW index statistics
• Skills count, curator status, auto-create status
• Active agents with live status indicators
• Routing tier breakdown with latency estimates
• Model provider status and configuration

File Structure

.effinai/
├── config.json          # Swarm config (topology, strategy, consensus)
├── usage.json           # Skill usage tracking (curator lifecycle)
├── agents/              # Custom agent definitions (your agents go here)
├── memory/
│   ├── entries.json     # Key-value memory store
│   └── patterns/        # Learned patterns with embeddings
│       └── index.json   # HNSW vector index
├── messages/            # Agent-to-agent communication (file-based)
├── swarm/
│   └── state.json       # Active swarm state
├── skills/              # Custom skills (your skills go here)
├── logs/                # Execution logs
├── multica.json         # Meta-orchestration config (if connected)
└── orchestrations/      # Multi-swarm orchestration tracking

Note: .effinai/agents/ and .effinai/skills/ are for your custom definitions. The 223 built-in agents and 98 built-in skills are bundled inside the package.

Configuration

effin config list                          # View all config
effin config set topology hierarchical     # Set value
effin config set maxAgents 12              # Adjust limits
effin config set learning.enabled true     # Toggle learning
effin config set consensus raft            # Set consensus mechanism

Environment Variables

# Model Providers
ANTHROPIC_API_KEY=sk-ant-...
OPENAI_API_KEY=sk-...
GOOGLE_API_KEY=...
XAI_API_KEY=...
KIMI_API_KEY=...
DEEPSEEK_API_KEY=...

# App Integrations
COMPOSIO_API_KEY=...

# Meta-Orchestration
MULTICA_URL=https://...
MULTICA_TOKEN=...

# Local Models
OLLAMA_HOST=http://localhost:11434
LM_STUDIO_HOST=http://localhost:1234

MCP Server Registration

# Register Effin.AI as MCP server for Claude Code
claude mcp add effin -- npx -y effinai@latest

# All 570 tools become available in Claude Code automatically
# MCP mode is auto-detected when stdin is piped (JSON-RPC 2.0 stdio)

Contributing

git clone https://github.com/rudycelekli/effinai-swarm.git
cd effinai-swarm/swarmmy
npm install
npm run build
npm link
effin --version

License

MIT License — Copyright (c) 2026 Effin.AI