> ## Documentation Index > Fetch the complete documentation index at: https://www.truefoundry.com/llms.txt > Use this file to discover all available pages before exploring further. # What TrueFoundry Adds on Top of TrueForge > How TrueFoundry turns the open-source TrueForge harness into a managed service: governance, observability, a managed runtime, and enterprise hosting. [TrueForge](https://trueforge.dev/introduction) is the open-source agent harness that runs the core agent loop — model calls, MCP tools, skills, sandboxing, approvals, context management, and session state. You can self-host it and wire up your own credentials, storage, and monitoring. TrueFoundry Agent Harness runs that same harness for you and layers on what a team needs to operate agents in production: centralized governance, end-to-end observability, a fully managed runtime, and enterprise hosting. No keys in agent code, plus RBAC, budgets, rate limits, and guardrails across models, MCP servers, and skills. Every agent session captured with cost, tokens, turns, tool calls, and end-to-end traces. Colocated sandbox, sensible defaults, and org-wide approval policy — nothing to wire up. Run as SaaS, self-hosted, or on-prem in your own cloud and region. ## Governance The single biggest difference between TrueFoundry Agent Harness and a self-hosted harness is that **no API keys or credentials are ever pasted into agent definitions**. Models, MCP servers, and skills are all managed through TrueFoundry's central control plane: * **Models** — Provider credentials live in the [AI Gateway](/docs/agent-platform/agent-harness/models). Agents reference model names. RBAC controls who can use which models. Budgets, rate limits, and guardrails are enforced at the AI Gateway. * **MCP Servers** — Authentication (OAuth tokens, API keys) lives in the [MCP Gateway](/docs/agent-platform/agent-harness/mcp-servers). Agents call tools by name. The AI Gateway handles credential injection, token refresh, and user delegation. * **Skills** — Published in the [Skills Registry](/docs/agent-platform/agent-harness/skills) with full versioning and RBAC. Agents pick from a governed catalog. Platform teams control what's available to whom. In Claude Managed Agents or LangSmith Managed Deep Agents, developers must register credentials (vault IDs, header arrays with bearer tokens) per agent or workspace. In TrueFoundry, platform teams configure access once and agent builders never handle secrets. ## Observability Because Agent Harness runs in the same gateway plane as model and MCP traffic, every agent session is captured with cost, tokens, turns, tool calls, and end-to-end traces — with no extra instrumentation in your agent code. This is one pane of glass across model, MCP, and agent traffic, and it inherits AI Gateway analytics, request logs, OpenTelemetry export, and Prometheus/Grafana. See [Observability](/docs/agent-platform/agent-harness/observability) for the Agent Sessions views, per-session metrics, and traces. ## Managed runtime The [harness capabilities](/docs/agent-platform/agent-harness/overview#capabilities-that-run-in-the-harness) themselves are part of open-source TrueForge. TrueFoundry runs them for you as a managed, governed service: * **Managed sandbox** — colocated with the harness for roughly 1ms execution latency, provisioned on demand, with lifecycle and cleanup handled for you. Secrets never enter the sandbox. * **Sensible defaults** — subagents, context compaction, large-result offloading, ask-user questions, and Generative UI are enabled out of the box; there is nothing to wire up. * **Governed approvals** — tools are flagged destructive once at the MCP Gateway, so the human-approval gate is enforced for every agent automatically instead of being configured per agent. * **Full observability** — every capability above shows up in [agent session traces](/docs/agent-platform/agent-harness/observability) with cost, tokens, and latency. ## Enterprise hosting TrueForge in local mode is meant for a single machine. TrueFoundry runs the harness as a shared, production service you can deploy as **SaaS** (globally distributed), **self-hosted**, or **on-prem** — in your own cloud and region, with authentication, RBAC, and the control plane included. ## Comparison with other harnesses Claude Managed Agents and LangSmith Managed Deep Agents are both strong hosted runtimes. The differences become clear when you look at how each platform handles builder experience, credentials, governance, observability, and deployment. | | TrueFoundry | Claude Managed Agents | LangSmith Managed Deep Agents | | --------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Builder experience & learning curve** | **No-code first** — non-developers ship agents from the Playground UI (pick a model, attach MCP servers and skills, write instructions, Save). Pro-code path via Python SDK and REST API for the same agent definition. Minutes to first working agent. | **Pro-code only** — define agents, environments, sessions as JSON payloads through the [API/SDK](https://platform.claude.com/docs/en/managed-agents/overview). No managed builder UI; developers write and maintain agent definitions in code. | **Pro-code only, API-first** — [private preview REST API](https://docs.langchain.com/langsmith/deploy-managed-deep-agent) consumed with `httpx`/`fetch`; SDK is "coming in a follow-up release." Agent definition lives in your repo (`AGENTS.md`, `skills/`, `subagents/`, `tools.json`) and is pushed via `POST/PATCH`. | | **MCP credentials** | Centralized in MCP Gateway with per-user OAuth, automatic refresh, and delegation. Users authenticate inline in chat (OAuth popup → continue). Admins rotate centrally — one update applies to every agent and user. | Per-user [Vaults](https://platform.claude.com/docs/en/managed-agents/vaults) you create programmatically and register per MCP server URL with `vault_ids` at session creation. No in-chat auth — developer acquires tokens externally. Rotate via `PATCH` per vault per user. | Static `headers` arrays registered via [`POST /v1/deepagents/mcp-servers`](https://docs.langchain.com/langsmith/deploy-managed-deep-agent#configure-mcp-servers). One credential set per workspace — no per-user isolation. OAuth-backed registration is planned but not yet available. | | **Tool approval & safety** | Tools flagged as destructive once at the MCP Gateway. Org-wide policy auto-enforced for every agent — agent builders don't configure anything. | Per-tool [`permission_policy`](https://platform.claude.com/docs/en/managed-agents/permission-policies) declared in each agent JSON (default `always_ask`). Forgetting to set it on a sensitive tool means it runs without confirmation. | Per-tool [`interrupt_config`](https://docs.langchain.com/langsmith/deploy-managed-deep-agent#define-the-agent) keyed by `{mcp_server_url}::{tool_name}`. Every agent must list every tool's interrupt preference; missed entries mean no approval gate. | | **Model access & governance** | Any provider via AI Gateway. Model-level RBAC, per-user/team budgets, rate limits, and pre/post-call guardrails (PII, content policies, custom). | Anthropic models only (`claude-opus-4-7`, `claude-sonnet-4-6`, ...). No model RBAC, budgets, or guardrails at the harness layer. | Any model via `{provider}:{model_id}` through [`init_chat_model`](https://docs.langchain.com/langsmith/deploy-managed-deep-agent#supported-models). No model RBAC, budgets, or guardrails in the managed runtime. | | **Observability** | Built-in end-to-end traces per agent run (LLM calls, tool calls, sandbox execs, subagents) with cost, tokens, and latency per step. Inherits AI Gateway analytics, request logs, OpenTelemetry export, and Prometheus/Grafana — one pane of glass across model, MCP, and agent traffic. | Server-side event history persisted per session and fetchable via API; SSE event stream during runs. No managed traces dashboard, cost analytics, or org-wide metrics surface at the harness layer. | [Traced in LangSmith](https://docs.langchain.com/langsmith/deploy-managed-deep-agent#inspect-the-result-in-langsmith) — inspect messages, tool calls, files, and subagent activity per run. Observability is scoped to LangSmith only; no cross-stack metrics, budgets, or alerting are part of the managed runtime. | | **Deployment** | SaaS (globally distributed), self-hosted, or on-prem — deployed in your own cloud and region. | [Managed Anthropic cloud](https://platform.claude.com/docs/en/managed-agents/overview) (US-only beta, EU post-GA) plus a separate self-hosted SDK path. No on-prem. | [Managed cloud](https://docs.langchain.com/langsmith/deploy-managed-deep-agent) (US-only private preview); self-host via `langgraph build`. No on-prem. | ## Next steps Create and test a managed agent from the TrueFoundry console. Track sessions, cost, and traces across every agent run.