Forge
Forge is an open-source, self-hosted alternative for visually building and shipping AI agents and workflows without giving up control of your infrastructure.
What is Forge?
Forge is an open-source, self-hosted alternative for visually building and shipping AI agents and workflows without giving up control of your infrastructure.
Overview of Forge
What it is
Forge is an open-source, self-hosted platform for visually building, testing, and shipping AI agents and workflows. It is built directly on the MIT-licensed LangChain + LangGraph framework and never depends on langgraph-api (Elastic 2.0) or LangSmith (commercial). Everything orchestrated runs on your own infrastructure; nothing is sent to a third-party orchestration service.
Key features
- Visual Agent Builder: Compose an Agent or a Deep Agent (planning + subagents) from a model, system prompt, tools, knowledge, Q&A, and a reorderable middleware stack.
- Visual Workflow Builder: Wire an entire app on a drag-and-drop canvas (React Flow) with typed, validated edges and nodes including agents, model calls, classifiers, tools, transforms, retrieval, human input, routers, loops, parallel fan-out/join, subworkflows, and triggers.
- Knowledge & RAG: Ground agents in your own data by adding pasted text, URLs, crawled sites, or uploaded files (.txt/.md/.csv/.json/.html/.pdf). Forge chunks, embeds (offline-capable by default), and stores them as vectors organized in folders.
- Tool Builder with Response Projection: Register REST, GraphQL, Code, SQL, MCP, or built-in tools and test them live. A JMESPath response projection trims bulky payloads before they reach the model, with an SSRF guard, optional retries, rate limits, and caching.
- Observability & Traces: Every run is captured as a collapsible span tree with model calls, tools, chains, sub-agents, latency, tokens, and cost. Traces can be exported to any OpenTelemetry collector.
- Guardrails, budgets & governance: Project-level Guardrails & Egress policy (PII redaction, blocked terms, network allow/deny list), budgets & quotas, versioning, per-project roles/RBAC, and an audit log.
Use cases
- Visually building and shipping AI agents and workflows without writing framework code.
- Grounding agents in proprietary data for retrieval-augmented generation (RAG).
- Deploying the same workflow through multiple surfaces: a single run API, an MCP server, email, or an embeddable web widget.
- Implementing human-in-the-loop approval and handoff to an Agent inbox.
- Running AI infrastructure entirely on your own servers with full control and observability.
Getting started
See the repository README for installation and quick-start instructions. The platform can run locally with SQLite and embedded Chroma without Docker, Postgres, or Redis. For production, swap to Postgres + pgvector, Redis, and a real secret store with configuration only.
License and pricing
License: MIT. Pricing: open-source and self-hosted. No proprietary core, no usage caps, no vendor lock-in.
Based on the project's public README and repository metadata.