Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/UnkleFunk/HouseMusicSwarm-/llms.txt

Use this file to discover all available pages before exploring further.

All OpenSwarm configuration lives in a single .env file at the root of the project. To get started, copy the template and fill in your keys: 
cp .env.example .env
The interactive setup wizard — launched by running openswarm (npm) or python onboard.py (local) — writes this file for you automatically. You can also edit it by hand at any time. Every key is optional except for one: you must supply an API key for at least one AI provider.
Never commit your .env file to version control. It contains secret API keys. The .gitignore in the OpenSwarm repo already excludes it, but double-check before pushing any fork or custom swarm.

​
AI provider (required — choose one)

OpenSwarm requires exactly one AI provider key to function. Choose the provider that best fits your use case and add the corresponding key to .env.
VariableProviderDefault modelGet your key
OPENAI_API_KEYOpenAIgpt-5.2platform.openai.com/api-keys
ANTHROPIC_API_KEYAnthropiclitellm/claude-sonnet-4-6console.anthropic.com/settings/keys
GOOGLE_API_KEYGoogle Geminilitellm/gemini/gemini-3-flashaistudio.google.com/app/apikey
Set the key for your chosen provider and leave the others blank. The setup wizard sets DEFAULT_MODEL automatically to match the provider you select.

​
Model selection

The DEFAULT_MODEL environment variable controls which model all agents use. It is set automatically by the setup wizard, but you can override it manually at any time. 
# OpenAI — model IDs contain no slash, passed directly to the API
DEFAULT_MODEL=gpt-5.2

# Anthropic — slash-prefixed strings are routed through LiteLLM automatically
DEFAULT_MODEL=litellm/claude-sonnet-4-6

# Google Gemini — also routed through LiteLLM
DEFAULT_MODEL=litellm/gemini/gemini-3-flash
How routing works: config.py reads DEFAULT_MODEL at startup. If the value contains a /, it is treated as a LiteLLM model string and wrapped in agency_swarm.LitellmModel. OpenAI model IDs (e.g. gpt-5.2, o3) contain no slash and are passed to the OpenAI API directly. This means you can use any LiteLLM-supported model by prefixing it with litellm/ — no other code changes required.
Run openswarm (or python onboard.py) again at any time to update your provider key, switch to a different model, or enable new add-ons through the interactive wizard. It detects existing values and only prompts for what needs to change.

​
Optional integrations

Composio extends the Virtual Assistant with access to over 10,000 external services including Gmail, Slack, GitHub, HubSpot, Google Calendar, Notion, Stripe, and more.
COMPOSIO_API_KEY=your_composio_api_key
COMPOSIO_USER_ID=your_composio_user_id
Both values are required. The user ID scopes which connected accounts the agent can access. Get your credentials at composio.dev.Without these keys, the Virtual Assistant operates without external integrations but still handles writing, planning, and local tasks.
The SEARCH_API_KEY enables web search, Google Scholar search, and product search across all agents — primarily used by the Deep Research agent.
SEARCH_API_KEY=your_searchapi_key
Get your key at searchapi.io. Without this key, the Deep Research agent cannot perform live web queries and will work from its training data only.
The GOOGLE_API_KEY enables Gemini-powered image generation and editing (Gemini 2.5 Flash Image, Gemini 3 Pro Image) across the Image Generation agent, Slides Agent, and Video Generation agent. It also unlocks Veo video generation in the Video Generation agent.
GOOGLE_API_KEY=your_google_ai_api_key
Get your key at aistudio.google.com/app/apikey.If you chose Google Gemini as your primary AI provider, this key is already set and also covers model inference — no separate key is needed.
The FAL_KEY enables Seedance 1.5 Pro video generation, video editing and clip combination, and background removal in the Image Generation agent.
FAL_KEY=your_fal_api_key
Get your key at fal.ai/dashboard/keys. Without this key, the Video Generation agent falls back to Sora (OpenAI) or Veo (Google) if those provider keys are available, and background removal in the Image agent is unavailable.
These keys enable the Slides Agent’s image search tool, which pulls high-quality stock photos into generated slide decks automatically.
PEXELS_API_KEY=your_pexels_api_key
PIXABAY_API_KEY=your_pixabay_api_key
UNSPLASH_ACCESS_KEY=your_unsplash_access_key
All three are optional and independent — the agent uses whichever keys are present. Get your keys at:Without any stock photo keys, the Slides Agent generates slides without embedded photos or uses AI-generated images if a GOOGLE_API_KEY or FAL_KEY is present.

​
Complete .env reference

The following is the full .env.example from the repository, showing every supported variable with inline documentation: 
# ─────────────────────────────────────────────
# OpenSwarm — environment configuration
# Copy this file to .env and fill in your keys.
# Run `python run.py` to launch the onboarding
# wizard, which fills this file automatically.
# ─────────────────────────────────────────────


# ── Provider (one required) ───────────────────

# OpenAI — set this if using OpenAI as your primary provider.
OPENAI_API_KEY=

# Anthropic Claude — set this if using Anthropic as your primary provider.
ANTHROPIC_API_KEY=

# Google Gemini — set this if using Google as your primary provider.
GOOGLE_API_KEY=


# ── Model selection ───────────────────────────

# Override the default model for all agents (set automatically by onboarding).
# OpenAI example:    DEFAULT_MODEL=gpt-5.2
# Anthropic example: DEFAULT_MODEL=litellm/claude-sonnet-4-6
# Google example:    DEFAULT_MODEL=litellm/gemini/gemini-3-flash
DEFAULT_MODEL=


# ── Optional: Composio ────────────────────────

# Composio unlocks 10,000+ external integrations (Gmail, Slack, Google Calendar,
# HubSpot, GitHub, and more) for the General Agent.
# Get your key and user ID at https://composio.dev
COMPOSIO_API_KEY=
COMPOSIO_USER_ID=


# ── Optional: Search ──────────────────────────

# Search API key — enables web search (WebSearchTool), Scholar search, and
# product search across agents.
# Get your key at https://www.searchapi.io
SEARCH_API_KEY=


# ── Optional: Images / Video ──────────────────

# Google AI / Gemini — used for image generation, editing, and composition
# (Gemini 2.5 Flash Image, Gemini 3 Pro Image) across Image Agent, Slides Agent,
# and Video Agent; also required for Veo video generation (Video Agent).
# Get your key at https://aistudio.google.com/app/apikey
GOOGLE_API_KEY=

# Pexels — stock photo search used by Slides Agent (ImageSearch tool).
# Get your key at https://www.pexels.com/api
PEXELS_API_KEY=

# Pixabay — stock photo search used by Slides Agent (ImageSearch tool).
# Get your key at https://pixabay.com/api/docs
PIXABAY_API_KEY=

# Unsplash — stock photo search used by Slides Agent (ImageSearch tool).
# Get your key at https://unsplash.com/developers
UNSPLASH_ACCESS_KEY=

# fal.ai — used for Seedance 1.5 Pro video generation, video editing (Video Agent),
# and background removal (Image Agent).
# Get your key at https://fal.ai/dashboard/keys
FAL_KEY=

Build docs developers (and LLMs) love

Get started for freeTalk to us