Skip to main content

bitHuman MCP server

[!IMPORTANT] Deprecated — use the bitHuman CLI instead. This server is now built into the bithuman CLI: run bithuman mcp. It exposes the same tools (identical names) plus local ones (version, doctor, inspect_model, list_showcase), so you install one tool. Migrate your MCP client config:

{ "mcpServers": { "bithuman": { "command": "bithuman", "args": ["mcp"] } } }

Install the CLI: brew install bithuman-product/bithuman/bithuman-cli (macOS) or the universal installer (macOS + Linux) — see the CLI README. This bithuman-mcp package will receive no further updates.

A Model Context Protocol server that exposes the bitHuman avatar platform as tools any MCP-capable AI agent can call — Claude Desktop, Claude Code, Cursor, and others.

It's a thin, fully-documented wrapper over the public REST API (https://api.bithuman.ai). Every tool maps to one documented endpoint; see the API docs and the OpenAPI spec.

Tools

The server exposes 22 tools, each mapping to one documented REST endpoint (plus get_platform_status, which reads the public status feed):

Tool What it does
get_platform_status Live platform + API status (public, no key).
validate_api_secret Check the API secret is valid (free).
get_credit_balance Current credits, plan, and minutes estimate.
get_usage Usage/metering history (paginated, date-filterable).
list_voices Built-in (M1–M5 / F1–F5) and custom TTS voices.
text_to_speech Synthesize speech → a WAV file.
generate_agent Create an avatar agent from a prompt / image / audio (image-only appearance; model+version select the engine, e.g. essence + v2 = essence-2).
get_agent_status Poll agent generation progress.
get_agent Fetch an existing agent's details.
list_agents List your agents, newest first (paginated).
update_agent_prompt Change an agent's system prompt.
delete_agent Permanently delete an agent you own.
agent_speak Make a live agent speak text in its active sessions.
add_agent_context Silently inject knowledge into a live agent.
get_dynamics List an agent's gesture animations.
generate_dynamics Generate new gestures (wave, nod, laugh, idle…).
create_embed_token Mint a 1-hour JWT to embed an agent on a website.
upload_file Upload an asset (URL or local file) → CDN URL.
create_webhook / list_webhooks / delete_webhook / test_webhook Manage signed event webhooks (agent.ready / agent.failed).

Setup

You need an API secret from the Developer Dashboard.

Published on PyPI as bithuman-mcp. The easiest way to run it is with uvx (recommended for MCP clients), or pip install bithuman-mcp.

BITHUMAN_API_SECRET=bh-... uvx bithuman-mcp

Use with Claude Desktop / Claude Code

Add it to your MCP client config. For Claude Code:

claude mcp add bithuman \
  -e BITHUMAN_API_SECRET=bh-your-secret \
  -- uvx bithuman-mcp

For Claude Desktop (claude_desktop_config.json) or any client that takes a JSON server block:

{
  "mcpServers": {
    "bithuman": {
      "command": "uvx",
      "args": ["bithuman-mcp"],
      "env": { "BITHUMAN_API_SECRET": "bh-your-secret" }
    }
  }
}

Configuration

Env var Default Purpose
BITHUMAN_API_SECRET (required) Your API secret. Never logged.
BITHUMAN_API_BASE https://api.bithuman.ai API origin.
BITHUMAN_MCP_TRANSPORT stdio stdio or streamable-http.
BITHUMAN_MCP_TIMEOUT 120 Per-request timeout (seconds).

Notes

  • Async work: generate_agent and generate_dynamics return immediately with a processing status. Poll get_agent_status / get_dynamics until ready (generation takes 2–5 minutes).
  • Credits: generate_agent (~250 credits) and text_to_speech consume credits. Check get_credit_balance first if cost matters.
  • Errors: non-2xx responses come back as a structured {error, status_code, body, hint} object. The error catalog is at https://docs.bithuman.ai/api/errors.

License

Apache-2.0.

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

bithuman_mcp-0.3.5.tar.gz (9.4 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

bithuman_mcp-0.3.5-py3-none-any.whl (10.6 kB view details)

Uploaded Python 3

File details

Details for the file bithuman_mcp-0.3.5.tar.gz.

File metadata

  • Download URL: bithuman_mcp-0.3.5.tar.gz
  • Upload date:
  • Size: 9.4 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/7.0.0 CPython/3.12.13

File hashes

Hashes for bithuman_mcp-0.3.5.tar.gz
Algorithm Hash digest
SHA256 d4c35defe1e6c04f03dee26e0af5a5d2982816e5e7a98b256895cf9c7ec28b3c
MD5 804b2155b2174b8b2acb6c6476d48487
BLAKE2b-256 0d5ad390991bc8d6a633bc5ad461f3362bc5ddece11a2d38812f7dff94bf4963

See more details on using hashes here.

File details

Details for the file bithuman_mcp-0.3.5-py3-none-any.whl.

File metadata

  • Download URL: bithuman_mcp-0.3.5-py3-none-any.whl
  • Upload date:
  • Size: 10.6 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/7.0.0 CPython/3.12.13

File hashes

Hashes for bithuman_mcp-0.3.5-py3-none-any.whl
Algorithm Hash digest
SHA256 9bbc79ed33ce0a27b0e4cc9a6481b53ca176ee08f4df3f884ddeb3b821a82989
MD5 9a027ac8801cf1149ab6cf7149d9b400
BLAKE2b-256 f9d7a48efd7196c5b4b3a2ec9fe3c1e3fe8c3777d9acf09de1e6f16dc065da44

See more details on using hashes here.

Release history Release notifications | RSS feed

This release

0.3.5 This release

2 files

0.3.4

2 files

0.3.3

2 files

0.3.2

2 files

0.3.1

2 files

0.3.0

2 files

0.2.1

2 files

0.2.0

2 files

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page