Skip to content

README.md

Latest commit

 

History

History
209 lines (154 loc) · 5.19 KB

README.md

File metadata and controls

209 lines (154 loc) · 5.19 KB
Raw

CrewForm CLI

Run CrewForm AI agents from the command line — scriptable, CI/CD-friendly, Ollama-first.

npx crewform run agent.json "Summarise the latest AI news"

Quick Start

# 1. Create an agent config
npx crewform init

# 2. Run it (Ollama, or set OPENAI_API_KEY / ANTHROPIC_API_KEY)
npx crewform run agent.json "Hello, world!"

# 3. Interactive chat
npx crewform chat agent.json

Installation

# Use directly (no install needed)
npx crewform

# Or install globally
npm install -g crewform

Commands

Local Execution

Command Description
crewform run <file> [prompt] Run an agent or team from a JSON config
crewform chat <file> Interactive chat session
crewform init Create a starter config file
crewform validate <file> Validate a config file
crewform tools List available built-in tools

Platform (API-Connected)

Command Description
crewform login Authenticate with your API key
crewform logout Remove saved credentials
crewform whoami Show authenticated user & workspace
crewform agents List agents in your workspace
crewform teams List teams in your workspace
crewform pull <id> Download agent/team config to local JSON
crewform push <id> [prompt] Dispatch a task to a remote agent/team

crewform run

# Basic usage
crewform run agent.json "Write a blog post about MCP"

# Read prompt from file
crewform run agent.json --input prompt.txt

# Save output to file
crewform run agent.json "Generate a report" --output report.md

# Pipe input/output
echo "Review this code" | crewform run agent.json
crewform run agent.json "Summarise" > summary.txt

# JSON output with metadata
crewform run agent.json "Hello" --json

# Quiet mode (no streaming, just result)
crewform run agent.json "Hello" --quiet

# Run a pipeline team
crewform run team.json "Research and write about GraphQL"

# Run with MCP servers
crewform run agent.json --mcp mcp-servers.json "Query my database"

crewform chat

Interactive REPL with conversation history:

crewform chat agent.json

Commands in chat:

  • /clear — Clear conversation history
  • /history — Show recent messages
  • /stats — Show token usage & cost
  • /exit — End session

crewform init

# Create agent config
crewform init

# Create pipeline team config
crewform init --team

# Custom name and model
crewform init --name "Code Reviewer" --model gpt-4o

Config Format

Inline Agent (simplest)

{
  "name": "My Agent",
  "model": "llama3.3",
  "system_prompt": "You are a helpful assistant.",
  "temperature": 0.7,
  "tools": ["web_search"]
}

CrewForm Export (from the web app)

The CLI also accepts the full crewform-export v1 format that you get when exporting an agent from the CrewForm web app.

Supported Providers

The CLI supports 16 LLM providers out of the box:

Provider Env Variable Notes
Ollama (none needed) Default, auto-detected
OpenAI OPENAI_API_KEY GPT-4o, o3, etc.
Anthropic ANTHROPIC_API_KEY Claude 4, etc.
Google GOOGLE_API_KEY Gemini 2.x
OpenRouter OPENROUTER_API_KEY 200+ models
Groq GROQ_API_KEY Fast inference
Mistral MISTRAL_API_KEY Codestral, etc.
Cohere COHERE_API_KEY Command R+
Together TOGETHER_API_KEY Open-source models
NVIDIA NVIDIA_API_KEY NIM models
HuggingFace HF_TOKEN Inference API
Venice VENICE_API_KEY Privacy-first
MiniMax MINIMAX_API_KEY
Moonshot MOONSHOT_API_KEY
Perplexity PERPLEXITY_API_KEY Sonar models

Available Tools

Tool Description Requires
web_search Search the web via Serper SERPER_API_KEY
http_request Make HTTP requests —
code_interpreter Run JavaScript in sandbox —
read_file Read file from URL —
grammar_check Check grammar/spelling —

MCP Server Support

Connect to MCP (Model Context Protocol) servers to give agents access to external tools:

crewform run agent.json --mcp mcp-servers.json "What tables are in my DB?"

mcp-servers.json:

[
  {
    "name": "my-db",
    "transport": "stdio",
    "command": "npx",
    "args": ["-y", "@modelcontextprotocol/server-postgres", "postgresql://..."]
  }
]

Supported transports: stdio, sse, streamable-http.

Platform Integration

Connect to the CrewForm web platform to manage agents remotely:

# Authenticate (get your API key from Settings → API Keys)
crewform login

# Browse your workspace
crewform agents
crewform teams

# Download an agent to run locally
crewform pull abc-123-uuid
crewform run my-agent.json "Hello"

# Dispatch work to a cloud agent/team
crewform push abc-123-uuid "Write a report on Q4 metrics"
crewform push abc-123-uuid --type team --wait "Research AI trends"

Credentials are saved to ~/.crewform/config.json. Self-hosted users can set --api-url.

License

AGPL-3.0-or-later · CrewForm