Agentic Flow

Community

Self-learning multi-agent orchestration for Claude Code.

How to install ↓ Compare GitHub source repository ↗

Category: Dev Tools #339 of 438
Stars: ★ 823 Very popular
Runtime: Node.js 18+
Credentials: No credential needed
Last commit: 17 days ago
Tools: 11

42FMRS · D

A developer-tools MCP service profile based solely on the supplied README. It is suited to Claude Code workflows involving agent routing, swarms, learning memory, and development automation. No server manifest was provided, so the transport and complete tool inventory cannot be confirmed; alpha status, permissions, and licensing should be verified before deployment.

Strongest · Documentation 11/20 Weakest · Security and permissions 5/20

Reliability

7/20

Security and permissions

5/20

Maintenance

9/20

Documentation

11/20

Setup experience

10/20

Why each score

Reliability 7/20: The README documents `npx agentic-flow mcp start` and claims a large MCP tool set, but the server manifest is unavailable and no source files, CI workflows, or committed tests verify MCP initialization, tool-list handshake, actual tool behavior, or error handling. Claims about 66 versus 80+ agents and alpha versus production-ready status are also inconsistent, so only limited plausibility is supported.
Security and permissions 5/20: The README indicates capabilities involving Bash, file edits, GitHub operations, workflows, and email-triggered automation; its generated permissions example includes `Bash(npx:*)` and broad MCP permissions. It provides no evidence of credential handling, least privilege, data boundaries, network disclosure, or confirmation for dangerous side effects. No explicit malware or secret theft is evidenced, so the 0-4 red line does not apply, but security controls remain materially unverified.
Maintenance 9/20: The repository is not archived, and the README supplies a GitHub issues channel, maintainer account, and an MIT license statement, plus a roadmap for future alpha releases. However, the supplied metadata says the license is unknown, and there is no evidence of commit cadence, release history, dependency updates, security response, or issue handling. The alpha status and README date of 2025-12-03 further limit confidence.
Documentation 11/20: The README offers extensive installation, initialization, commands, architecture, examples, environment variables, and benchmark claims. Deductions are warranted because there is no reviewable manifest, authentication guidance, complete tool parameter/limitation coverage, cost disclosure, or troubleshooting material. It also contains unresolved inconsistencies such as 66 versus 80+ agents, alpha versus production-ready labeling, and performance/testing claims not verifiable from the supplied source.
Setup experience 10/20: The README specifies Node.js 18+, npm, `npx agentic-flow init`, an MCP start command, and a Claude Code configuration example, making the nominal path fairly short. Without source files, client-connection validation, CI, or committed tests proving the handshake and platform behavior, and with an alpha package, optional native NAPI rebuild, and generated hooks, setup remains potentially fragile and cannot score near the top.

Static review · not runListed 2026-07-29

Read the FMRS scoring method →

Fit and risk

What it can accessReads local filesWrites / deletes local filesRuns commands or codeUses the networkConnects to a database

Best for

Developers familiar with Claude Code, the Claude Agent SDK, or multi-agent workflows
Teams needing self-learning routing, swarm coordination, and project-level code analysis
Users building agent applications with the TypeScript API

Not for

Users who only need one simple MCP tool
Environments that do not permit project-level command execution or codebase access
Production environments requiring a verified stable release rather than an alpha version

Required permissions

Access to project files and file structure for initialization, pretraining, code analysis, and edit-related hooks
Execution of npm, npx, build, test, and other user-triggered shell commands
Git history access during pretraining, plus repository or pull-request access when GitHub tools are used
Permission for the MCP client to start the local npx-configured service

Risks and side effects

The README calls the project production-ready, but installation examples and the roadmap still use alpha releases; stability should be verified
Self-learning memory may store project-related task inputs, outputs, scores, critiques, and token data
Agents and background workers may perform broad code analysis, command execution, or automation
Performance, accuracy, and cost-saving figures are project-reported claims and are not independently verified by the supplied material
Repository metadata lists the license as unknown, while the README claims MIT; verify the repository LICENSE before use

Setup

Before you start

Runtime：Node.js 18+

Other optional settings (5)

AGENTIC_FLOW_INTELLIGENCE optional Toggle to enable the intelligence learning layer (true/false); optional.

AGENTIC_FLOW_LEARNING_RATE optional Q-learning rate (0.0-1.0), default 0.1.

AGENTIC_FLOW_MEMORY_BACKEND optional Memory backend: agentdb, sqlite, or memory.

AGENTIC_FLOW_WORKERS_ENABLED optional Enable the background workers system; optional.

AGENTIC_FLOW_MAX_WORKERS optional Maximum number of background workers; optional.

Requires Node.js 18+ and npm 8+. Install with npm install agentic-flow@alpha, then start the MCP service with npx agentic-flow@alpha mcp start. You can also run npx agentic-flow init to initialize a project, followed by hooks pretrain and hooks build-agents. The README provides a Claude Code mcpServers configuration example.

.mcp.json

{"mcpServers":{"agentic-flow":{"command":"npx","args":["agentic-flow@alpha","mcp","start"]}}}

Shown for Claude Code. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json

{
  "servers": {
    "agentic-flow": {
      "command": "npx",
      "args": [
        "agentic-flow@alpha",
        "mcp",
        "start"
      ]
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal

claude mcp add agentic-flow -- npx agentic-flow@alpha mcp start

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

The agentic-flow tools (swarm_init, agent_spawn, task_orchestrate, etc.) should appear in your MCP client's tool list; invoking benchmark_run and getting benchmark results confirms the connection works.

Troubleshooting

Confirm Node.js 18 or newer and npm 8 or newer
Confirm agentic-flow@alpha is installed and run commands from the repository root or target project
Check the generated .claude/settings.json and its mcpServers command and arguments
For source installation, run npm run build or npm test to check the setup
For native acceleration, verify that @ruvector/attention rebuilt successfully; the README also describes WASM and JavaScript fallbacks

Things to try

Once connected, you can ask your AI assistant things like:

Initialize a swarm and spawn coder, tester, and reviewer agents to implement user login
Analyze my GitHub repository's architecture and identify performance bottlenecks
Check recent neural patterns, then run neural_train focused on code tasks
Run a benchmark and report token usage

Tools 11

swarm_init writes

Initializes multi-agent swarm coordination.

agent_spawn writes

Creates an agent.

task_orchestrate writes

Orchestrates task execution.

memory_usage read-only

Views or manages memory usage.

neural_train writes

Performs neural learning or training operations.

neural_patterns read-only

Processes neural patterns.

github_repo_analyze read-only

Analyzes a GitHub repository.

github_pr_manage writes

Manages or reviews a GitHub pull request.

Show 3 more tools

benchmark_run read-only

Runs performance benchmarks.

bottleneck_analyze read-only

Analyzes performance bottlenecks.

token_usage read-only

Reports token usage.

Use cases

Task routing and software development in Claude Code projects

Parallel code review, testing, and CI/CD automation

Enterprise document processing, search, and workflow automation

Research analysis and multi-agent collaboration

Supported clients

Claude CodePartial support

Listed from the project's documentation, not tested by this site.

Overview

Agentic-Flow v2 is a Node.js and TypeScript framework for AI-agent orchestration. Its README claims 66 self-learning agents, 213 MCP tools, multi-agent swarms, memory and vector search, task routing, background workers, and Claude Code integration.

Similar servers

Context7 80 · B

Upstash's official server providing up-to-date third-party library docs for AI coding assistants

★ 63.3k · Tools 2 Compare with this →

NPM Sentinel MCP Server 79 · B

AI-powered NPM package analysis MCP server

★ 18 · Tools 19 Compare with this →

MartinLoop MCP Server 79 · B

An independent control layer that gives AI coding agents budgets, brakes, and receipts

★ 413 Compare with this →

GitHub MCP Server 76 · B

GitHub's official MCP server for managing repos, issues, PRs, and workflows via natural language

★ 33.7k · Tools 15 Compare with this →

Compare with other MCP servers

Source revision a2f6c1cfa709 Data synced 2026-10-19 Read the FMRS scoring method