Ross ROSS = Recommend OSS · open-source software intelligence for agents

truefoundry/trueforge

The open-source agent harness - the runtime layer that turns an LLM into a working agent. observed · 2026-08-28

github.com/truefoundry/trueforge · homepage · TypeScript · MIT (permissive) observed · 2026-08-28

Health v2 · maintenance only

79/100

  • Activity 99
  • Release rhythm 98
  • Longevity 2

Flags: young

How is this computed?

round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10) — computed 2026-09-03. Adoption (stars, forks) is never an input.

  • gap_med: 0
  • age_days: 41
  • days_rel: 14
  • days_push: 7
  • n_releases_24m: 26

Full methodology

Adoption not part of the score

4552 stars · 289 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

TrueForge is an open-source agent harness that provides the runtime layer turning an LLM into a working agent, with MCP tools, skills, sandboxing, human approvals, and subagents built in. It ships as a core server running the agent loop, a TypeScript SDK over an HTTP API, and a React chat UI SDK, deployable locally via npx or hosted with Postgres and Redis.

Use cases

  • run a self-hosted agent runtime with MCP tool support
  • build a custom chat agent with streaming and tool calls
  • add human-in-the-loop approvals to sensitive agent actions
  • run long-lived agent sessions that survive reconnects and restarts
  • execute agent code and shell commands in an isolated sandbox
  • embed an agent chat UI into a React app
  • manage context on long agent tasks with compaction and subagents

When to choose

  • you need a production-grade agent runtime with sandboxing, approvals, and session persistence
  • you want MCP tool integration and context engineering out of the box
  • you want both a hosted team deployment and a local personal mode
  • you prefer TypeScript/Node.js and bring-your-own model providers

When to avoid

  • you only need a simple LLM API wrapper or prompt library
  • you want the entire agent loop to run inside a VM rather than a server-side sandbox
  • you need a Python-first agent framework
  • local mode is unsuitable for internet-facing use since it has no login by default

Facets

framework · maturity active

agent-framework mcp chat-interface http-server api-framework llm-inference large-language-models developer-tools self-hosted self-hosted cross-platform agent-harness mcp-tools human-in-the-loop context-engineering subagents tool-approval chat-ui react-sdk session-persistence sandbox ai-agents nodejs typescript docker kubernetes

6 sources

Member repositories

RepositoryRoleHealth v2
truefoundry/trueforgemain79

For agents

markdown · JSON · MCP: product_card(name="truefoundry/trueforge")

Data as of 2026-08-30T08:39:29.467469+00:00 · Report a problem