Langflow logo

Langflow

Active

Open-source visual builder for AI agents and MCP servers.

Conditional

Langflow is an MIT-licensed visual builder for AI agents and MCP servers that you can self-host with Desktop, Docker or pip. The deciding caveat: per The Hacker News, releases 1.0.0 through 1.12.8 contain critical flaws, including two rated 9.8, fixed in 1.13.0, so run 1.13.0 or later and set a superuser password instead of relying on auto-login. It suits developers who want a self-hosted flow builder and will patch promptly. We have not tested it.

AI-Powered

Users draw graphs of discrete components; an agent component can call tools, but the user-defined graph, not one AI loop, drives the work.

Is it right for you?

Good for

  • Building agent and RAG flows visually. The site describes a drag-and-drop editor with a component library and pre-built flows.
  • Turning a flow into an MCP server. The README says: "Deploy as an MCP server and turn your flows into tools for MCP clients".
  • Self-hosting under a permissive license. The repository is MIT licensed and installs via pip or Docker.

Not good for

  • Running a release from 1.0.0 through 1.12.8 on a reachable host. The Hacker News lists four flaws fixed in 1.13.0, two of them critical at CVSS 9.8, one high (8.8) and one medium (6.5).
  • Expecting a quiet security record. Five of the ten advisories we read on GitHub's advisory page are rated Critical, published between March 16 and September 10, 2026.
  • Teams that need a published hosted price. The pages we read list none.
  • Langflow Desktop users who need the Shareable Playground or Voice Mode. The install docs say Desktop lacks both.

Our experience

We haven't tested Langflow ourselves. This profile is based on public documentation, user reviews, and community feedback.

Pricing

Open Source
Self-hosted (OSS)Free
  • MIT license; Desktop, Docker or pip; you pay for your own hosting and model APIs
View full pricing

Workflow recipes

Expose a flow as an MCP tool

Build a flow in Langflow and let an MCP client call it.

  1. Install Langflow 1.13.0 or later (Desktop, Docker or the Python package).
  2. Set LANGFLOW_SUPERUSER_PASSWORD before starting Langflow, as the README directs.
  3. Build the flow in the visual editor.
  4. Deploy the flow as an MCP server.
  5. Connect your MCP client to it.

No verdict changes yet

The clock starts day one. Changes land here as our verdict evolves.

Verification log

  • Content· Updated

    Automated agent

    Security wording corrected: four flaws in 1.0.0 through 1.12.8 (CVE-2026-93680 and CVE-2026-110186 at CVSS 9.8, CVE-2026-110193 at 8.8, CVE-2026-110192 at 6.5), fixed in 1.13.0; releases page lists v1.13.1 and v1.13.2. Description, strengths and sources refreshed. Secondary source only; the GitHub advisories page does not list these CVEs, so no verdict change.

How we evaluate

Want this running in your business?

We tested this tool and we build with tools like it every day. Tell us the workflow and we will set it up, integrate it and hand it over working.

The conversation runs on Gnosari, one of the tools in this directory. A real conversation, not a sales script.