Langfuse Agent

API | MCP | Agent

Version: 2.3.0

Documentation — Installation, deployment, usage across the API, CLI, and MCP interfaces, and guidance for provisioning the Langfuse platform are maintained in the official documentation.

Overview

Langfuse Agent is a production-grade Agent and Model Context Protocol (MCP) server designed to interface directly with the Langfuse LLM Engineering and Observability platform. It enables agentic models to query, create, and manage observability traces, datasets, prompt templates, and system configurations.

Key Features

• Consolidated Action-Routed MCP Tools: Minimizes token overhead and eliminates tool bloat in LLM contexts by grouping 80+ methods into 4 optimized, togglable tool modules.
• Enterprise-Grade Security: Complete support for Eunomia policies, OIDC token delegation, and granular execution context tracking.
• Integrated Graph Agent: Built-in Pydantic AI agent supporting the Agent Control Protocol (ACP) and standard Web interfaces (AG-UI).
• Optional Telemetry & Tracing: OTLP export and Langfuse instrumentation enable only when their runtime configuration is present.

CLI or API

This agent wraps the Langfuse API. The operator can interact with it programmatically or via its integrated execution entrypoints.

Detailed instructions on how to use the underlying API wrappers, extended schema bindings, and developer SDK references are maintained in docs/index.md.

MCP

This server use dynamic Action-Routed tools to optimize token overhead and maximize IDE compatibility.

Available MCP Tools

Auto-generated — do not edit between the markers below.

Condensed action-routed tools (MCP_TOOL_MODE=condensed or both)

Verbose 1:1 API-mapped tools (MCP_TOOL_MODE=verbose or both)

5 action-routed tool(s) · 81 verbose 1:1 tool(s). MCP_TOOL_MODE selects the surface (intent default · condensed action-routed · verbose 1:1 · both). Auto-generated — do not edit.

Detailed tool schemas, parameter shapes, and validation constraints are preserved in docs/usage.md.

Dynamic Tool Selection & Visibility

This MCP server supports dynamic toolset selection and visibility filtering at runtime. This allows the operator to restrict the set of exposed tools to prevent blowing up the LLM's context window.

The operator can configure tool filtering via multiple input channels:

• CLI Arguments: Pass --tools or --toolsets (or their disabled counterparts --disabled-tools and --disabled-toolsets) during startup.
• Environment Variables: Define standard environment variables:
  • MCP_ENABLED_TOOLS / MCP_DISABLED_TOOLS
  • MCP_ENABLED_TAGS / MCP_DISABLED_TAGS
• HTTP SSE Request Headers: Pass custom headers during transport initialization:
  • x-mcp-enabled-tools / x-mcp-disabled-tools
  • x-mcp-enabled-tags / x-mcp-disabled-tags
• HTTP SSE Request Query Parameters: Append query parameters directly to the operator's transport connection URL:
  • ?tools=tool1,tool2
  • ?tags=tag1

When query strings or parameters are provided, an LLM-free Knowledge Graph resolution layer (using DynamicToolOrchestrator) matches query intents against known tool tags, names, or descriptions, with safe fallback and automated 24-hour background cache refreshing.

MCP Configuration Examples

Install the connector-focused [mcp] extra. Examples use langfuse-agent[mcp] to add FastMCP / FastAPI through agent-connector-sdk[mcp]; the required Agent Utilities core still carries epistemic-graph[full]. The [agent-runtime] extra additionally enables model orchestration.

stdio Transport (local IDEs — Cursor, Claude Desktop, VS Code)

{
  "mcpServers": {
    "langfuse-mcp": {
      "command": "uvx",
      "args": [
        "--from",
        "langfuse-agent[mcp]",
        "langfuse-mcp"
      ],
      "env": {
        "MCP_TOOL_MODE": "intent",
        "LANGFUSE_DATASETSTOOL": "True",
        "LANGFUSE_KGTOOL": "True",
        "LANGFUSE_MANAGEMENTTOOL": "True",
        "LANGFUSE_OBSERVABILITYTOOL": "True",
        "LANGFUSE_PROMPTS_MODELSTOOL": "True"
      }
    }
  }
}

Runtime references require an alias-aware launcher such as Arbora. Other launchers must omit those entries and inject the resolved values through their own runtime secret boundary.

Streamable-HTTP Transport (networked / production)

{
  "mcpServers": {
    "langfuse-mcp": {
      "command": "uvx",
      "args": [
        "--from",
        "langfuse-agent[mcp]",
        "langfuse-mcp",
        "--transport",
        "streamable-http",
        "--port",
        "8000"
      ],
      "env": {
        "TRANSPORT": "streamable-http",
        "HOST": "127.0.0.1",
        "PORT": "8000",
        "MCP_TOOL_MODE": "intent",
        "LANGFUSE_DATASETSTOOL": "True",
        "LANGFUSE_KGTOOL": "True",
        "LANGFUSE_MANAGEMENTTOOL": "True",
        "LANGFUSE_OBSERVABILITYTOOL": "True",
        "LANGFUSE_PROMPTS_MODELSTOOL": "True"
      }
    }
  }
}

Alternatively, connect to a pre-deployed Streamable-HTTP instance by url:

{
  "mcpServers": {
    "langfuse-mcp": {
      "url": "http://localhost:8000/langfuse-mcp/mcp"
    }
  }
}

Run a reviewed container image as a least-privilege stdio child (no listener or published port):

docker run -i --rm \
  --read-only \
  --cap-drop=ALL \
  --security-opt=no-new-privileges \
  --pids-limit=256 \
  --tmpfs /tmp:rw,noexec,nosuid,nodev,size=64m \
  -e TRANSPORT=stdio \
  -e MCP_TOOL_MODE=intent \
  -e LANGFUSE_DATASETSTOOL=True \
  -e LANGFUSE_KGTOOL=True \
  -e LANGFUSE_MANAGEMENTTOOL=True \
  -e LANGFUSE_OBSERVABILITYTOOL=True \
  -e LANGFUSE_PROMPTS_MODELSTOOL=True \
  registry.example.invalid/langfuse-agent@sha256:<digest> langfuse-mcp

For containerized network HTTP, supply an authenticated TLS ingress (or direct server TLS), exact MCP_ALLOWED_HOSTS, and an exact trusted-proxy CIDR policy through the operator-owned deployment profile. The generator does not emit an unauthenticated non-loopback listener.

Auto-generated from the code-read env surface (MCP_TOOL_MODE + package vars) — do not edit.

Additional Deployment Options

langfuse-agent can run as a local stdio process or container, or behind a remote network boundary. The Deployment guide carries the detailed transport contract.

• Local container — launch a reviewed immutable image as a least-privilege stdio child with no listener or published port.
• Remote URL — connect through an operator-supplied authenticated HTTPS ingress. Keep its URL, outbound identity references, trust profile, and exact MCP_ALLOWED_HOSTS in AgentConfig.

Agent

This repository features a fully integrated Pydantic AI Graph Agent. It communicates over the Agent Control Protocol (ACP) and interacts smoothly with the Agent Web UI (AG-UI) and Terminal interface.

Running the Agent CLI

To start the interactive command-line agent, first let the process supervisor inject the Langfuse credentials and any model-provider secret. Then run:

langfuse-agent --provider openai --model-id gpt-4o

Docker Compose Orchestration

The checked-in docker/agent.compose.yml runs the MCP and agent services with restricted containers and loopback-published ports. Provide pinned image digests and let the process supervisor inject materialized child credentials at runtime; never add them to the Compose file or a checked-in env file. Arbora deployments keep only the corresponding *_REF settings in AgentConfig.

Detailed graph node architecture explanations, custom skill configurations, and agentic trace guides are available in docs/deployment.md.

Security & Governance

Built directly upon the enterprise-ready agent-connector-sdk core, standard security parameters are fully supported:

Access Control & Policy Enforcement

• Eunomia Policies: Fine-grained, policy-driven tool authorization. Supports none, local embedded (mcp_policies.json), or centralized remote modes.
• OIDC Token Delegation: Compliant with RFC 8693 token exchange for flowing authenticating user credentials from Web UI / ACP → Agent → MCP.
• Scoped Credentials: Execution context runs restricted to the specific caller identity.

Runtime Security Grid

Configuration & Environment Variables

Arbora configuration uses AgentConfig and secret references. Do not store credential values in .env files or MCP catalogs.

Core API & Credentials

Server Configuration

Agent Customization

Tool Toggle Switches

Individual tool modules can be enabled or disabled to minimize client context size. These names match the authoritative "Toggle Env Var" column in the Available MCP Tools table above:

• LANGFUSE_OBSERVABILITYTOOL (Default: True): Toggles observation/tracing tools.
• LANGFUSE_DATASETSTOOL (Default: True): Toggles datasets and annotation queue tools.
• LANGFUSE_PROMPTS_MODELSTOOL (Default: True): Toggles prompt template and model connectivity tools.
• LANGFUSE_MANAGEMENTTOOL (Default: True): Toggles comments, SCIM, and project management tools.

Installation

Pick the extra that matches what the operator want to run:

# Connector-focused MCP server (includes the shared graph engine)
uv pip install "langfuse-agent[mcp]"

# Agent runtime (adds model orchestration to the shared graph engine)
uv pip install "langfuse-agent[agent]"

# Everything (development)
uv pip install "langfuse-agent[all]"      # or: python -m pip install "langfuse-agent[all]"

Container image targets

One multi-stage docker/Dockerfile builds two published profiles and one exact, offline release profile selected by --target:

docker build --target mcp   -t langfuse-agent:mcp docker/
docker build --target agent -t langfuse-agent:agent docker/

Promote and deploy only an operator-reviewed immutable image digest.

Knowledge-graph database (epistemic-graph)

Both [mcp] and [agent] carry the epistemic-graph engine through the required Agent Utilities core dependency (epistemic-graph[full], including its folded numeric kernel). The [mcp] extra keeps the server connector-focused; [agent] also enables model orchestration. Local deployments can use the bundled engine. For production or shared state, run epistemic-graph as a dedicated database service and configure the runtime to use it. Deployment recipes (single-node + Raft HA), connection configuration, and architecture diagrams are documented in the epistemic-graph deployment guide.

Documentation

The complete documentation is published as the official documentation site and is the recommended reference for installation, deployment, and day-to-day operation.

Repository Owners

Contribute

Contributions are welcome! Please ensure code quality by running local checks before submitting pull requests:

• Format code using ruff format .
• Lint code using ruff check .
• Validate type-safety with mypy .
• Ran test suites using pytest

Deploy with agent-connector-sdk-deployment

Provision this package with the consolidated agent-connector-sdk-deployment workflow. It selects an installed-package, editable-source, or immutable-container path; records only runtime secret and TLS-profile references in AgentConfig; and runs doctor, registration, policy, observability, and rollback gates. Ask your agent to "deploy langfuse-agent with agent-connector-sdk-deployment".

The repository embeds no deployment profile, credential value, certificate path, or environment-specific endpoint. Supply those at runtime through AgentConfig and the configured secret provider.

Environment Variables

Package environment variables

Inherited agent-connector-sdk variables (apply to every connector)

33 package + 15 inherited variable(s). Auto-generated from .env.example + the shared agent-connector-sdk set — do not edit.

Governed capability contract

This package ships a compact canonical skill surface with specialist procedures kept as referenced workflows. The current MCP tools, skill metadata, connector_manifest.yml, ontology, mappings, shapes, fixtures, migrations, tool-schema fingerprints, and certification metadata form one versioned capability contract. Validate them together; do not rely on stale tool names or historical per-task skill wrappers.

Runtime endpoints, credentials, certificate trust, tenant identity, retention, and observability policy are deployment inputs and are never packaged values. See Configuration, trust, and privacy before enabling a network transport, connector ingestion, Arbora delegation, or trace export.