# FlowStack SDK
**From local agent to production in 5 minutes.**
FlowStack is the simplest way to build and deploy AI agents. Write Python tools, test locally, deploy instantly. No infrastructure, no DevOps, no complexity.
## 🚀 Why FlowStack?
**Think Vercel for AI agents.** You focus on agent logic, we handle everything else.
- ✅ **Write tools in Python** - Natural @agent.tool decorator
- ✅ **Deploy in one command** - `agent.deploy()` and you're live
- ✅ **Instant production API** - Get HTTPS endpoints immediately
- ✅ **Built-in memory** - DataVault for persistent agent state
- ✅ **Multi-provider support** - Switch between OpenAI, Claude, Llama seamlessly
## 🎯 Perfect For
- **🥷 Indie developers** - Build weekend projects that scale
- **🚀 Startups** - Rapid prototyping without infrastructure overhead
- **🤖 Automation enthusiasts** - Replace Zapier with custom AI logic
## ⚡ Quick Start
### 1. Install & Deploy in 5 Minutes
```bash
# Install
pip install flowstack
# Get your API key from flowstack.fun
export FLOWSTACK_API_KEY="fs_..."
```
### 2. Build Your First Agent
```python
from flowstack import Agent
# Create agent with tools
agent = Agent("customer-helper")
@agent.tool
def lookup_order(order_id: str) -> dict:
"""Look up customer order status"""
# Your business logic here
return {"status": "shipped", "tracking": "UPS123456789"}
@agent.tool
def update_address(order_id: str, new_address: str) -> str:
"""Update shipping address for an order"""
# Your business logic here
return f"Address updated for order {order_id}"
# Test locally
response = agent.chat("What's the status of order #12345?")
print(response) # Agent calls lookup_order() automatically
```
### 3. Deploy to Production
```python
# Deploy with one command
endpoint = agent.deploy()
print(f"🎉 Your agent is live at: {endpoint}")
# Now available as REST API:
# POST https://your-agent.flowstack.fun/chat
# {"message": "What's the status of order #12345?"}
```
## 🧠 Persistent Memory with DataVault
Agents remember context across conversations:
```python
@agent.tool
def save_preference(user_id: str, preference: str) -> str:
"""Save user preference"""
agent.vault.store('preferences', {
'user_id': user_id,
'preference': preference
})
return "Preference saved!"
@agent.tool
def get_recommendations(user_id: str) -> list:
"""Get personalized recommendations"""
prefs = agent.vault.query('preferences', {'user_id': user_id})
# Use preferences to customize recommendations
return ["item1", "item2", "item3"]
```
## 🔄 Multi-Provider Flexibility
Switch AI providers without changing code:
```python
# Use managed Bedrock (included in pricing)
agent = Agent("my-agent", model="claude-3-sonnet")
# Or bring your own OpenAI key
agent = Agent("my-agent",
provider="openai",
model="gpt-4o",
byok={"api_key": "sk-your-key"}
)
# Same tools, same deployment, different AI provider
```
## 📊 Real-World Examples
### Slack Customer Support Bot
```python
agent = Agent("support-bot")
@agent.tool
def search_knowledge_base(query: str) -> str:
# Search your docs/FAQ
return search_results
@agent.tool
def create_ticket(issue: str, priority: str) -> str:
# Create Zendesk/Jira ticket
return ticket_id
# Deploy and connect to Slack webhook
endpoint = agent.deploy()
```
### E-commerce Order Assistant
```python
agent = Agent("order-assistant")
@agent.tool
def check_inventory(product_id: str) -> dict:
# Check your inventory system
return {"in_stock": True, "quantity": 50}
@agent.tool
def process_return(order_id: str, reason: str) -> str:
# Handle return logic
return "Return processed"
# Deploy and embed in your website
endpoint = agent.deploy()
```
### Content Creation Workflow
```python
agent = Agent("content-creator")
@agent.tool
def research_topic(topic: str) -> str:
# Research from multiple sources
return research_summary
@agent.tool
def publish_to_cms(title: str, content: str) -> str:
# Publish to WordPress/Contentful
return "Published successfully"
# Deploy and trigger via webhook
endpoint = agent.deploy()
```
## 💰 Simple Pricing
**Session-based pricing.** One API call = one session, regardless of message count.
- **Free**: 25 sessions/month
- **Hobbyist**: $15/month, 200 sessions
- **Starter**: $50/month, 1,000 sessions
- **Professional**: $200/month, 5,000 sessions
- **Enterprise**: Custom pricing, 25,000+ sessions
## 📚 Learn More
- **📖 [Documentation](https://docs.flowstack.fun/)** - Complete guides and examples
- **🚀 [5-Minute Tutorial](https://docs.flowstack.fun/quickstart/)** - Get started immediately
- **🧠 [DataVault Guide](https://docs.flowstack.fun/datavault/)** - Persistent agent memory
- **🍳 [Recipe Collection](https://docs.flowstack.fun/recipes/chatbot/)** - Real-world examples
## 🛠️ Development
```bash
# Local development
git clone https://github.com/flowstack-fun/flowstack.git
cd flowstack
pip install -e .
# Run tests
pytest tests/
# Build docs
mkdocs serve
```
## 🤝 Support
- **🐛 Issues**: [GitHub Issues](https://github.com/flowstack-fun/flowstack/issues)
- **💬 Community**: [Discord](https://discord.gg/flowstack)
- **📧 Email**: [support@flowstack.fun](mailto:support@flowstack.fun)
- **🐦 Updates**: [@flowstack](https://twitter.com/flowstack)
---
**Built for developers who ship fast.** 🚢
*Stop building infrastructure. Start building agents.*
Raw data
{
"_id": null,
"home_page": "https://github.com/flowstack-fun/flowstack",
"name": "flowstack",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.8",
"maintainer_email": null,
"keywords": "ai, agents, llm, openai, anthropic, bedrock, aws",
"author": "FlowStack",
"author_email": "hello@flowstack.ai",
"download_url": "https://files.pythonhosted.org/packages/0d/fe/5137281f5bf3922030ce51c277dd77cf252b35f1473373aafc0a76cf78f2/flowstack-1.0.0.tar.gz",
"platform": null,
"description": "# FlowStack SDK\n\n**From local agent to production in 5 minutes.**\n\nFlowStack is the simplest way to build and deploy AI agents. Write Python tools, test locally, deploy instantly. No infrastructure, no DevOps, no complexity.\n\n## \ud83d\ude80 Why FlowStack?\n\n**Think Vercel for AI agents.** You focus on agent logic, we handle everything else.\n\n- \u2705 **Write tools in Python** - Natural @agent.tool decorator\n- \u2705 **Deploy in one command** - `agent.deploy()` and you're live\n- \u2705 **Instant production API** - Get HTTPS endpoints immediately \n- \u2705 **Built-in memory** - DataVault for persistent agent state\n- \u2705 **Multi-provider support** - Switch between OpenAI, Claude, Llama seamlessly\n\n## \ud83c\udfaf Perfect For\n\n- **\ud83e\udd77 Indie developers** - Build weekend projects that scale\n- **\ud83d\ude80 Startups** - Rapid prototyping without infrastructure overhead\n- **\ud83e\udd16 Automation enthusiasts** - Replace Zapier with custom AI logic\n\n## \u26a1 Quick Start\n\n### 1. Install & Deploy in 5 Minutes\n\n```bash\n# Install\npip install flowstack\n\n# Get your API key from flowstack.fun\nexport FLOWSTACK_API_KEY=\"fs_...\"\n```\n\n### 2. Build Your First Agent\n\n```python\nfrom flowstack import Agent\n\n# Create agent with tools\nagent = Agent(\"customer-helper\")\n\n@agent.tool\ndef lookup_order(order_id: str) -> dict:\n \"\"\"Look up customer order status\"\"\"\n # Your business logic here\n return {\"status\": \"shipped\", \"tracking\": \"UPS123456789\"}\n\n@agent.tool \ndef update_address(order_id: str, new_address: str) -> str:\n \"\"\"Update shipping address for an order\"\"\"\n # Your business logic here\n return f\"Address updated for order {order_id}\"\n\n# Test locally\nresponse = agent.chat(\"What's the status of order #12345?\")\nprint(response) # Agent calls lookup_order() automatically\n```\n\n### 3. Deploy to Production\n\n```python\n# Deploy with one command\nendpoint = agent.deploy()\nprint(f\"\ud83c\udf89 Your agent is live at: {endpoint}\")\n\n# Now available as REST API:\n# POST https://your-agent.flowstack.fun/chat\n# {\"message\": \"What's the status of order #12345?\"}\n```\n\n## \ud83e\udde0 Persistent Memory with DataVault\n\nAgents remember context across conversations:\n\n```python\n@agent.tool\ndef save_preference(user_id: str, preference: str) -> str:\n \"\"\"Save user preference\"\"\"\n agent.vault.store('preferences', {\n 'user_id': user_id, \n 'preference': preference\n })\n return \"Preference saved!\"\n\n@agent.tool\ndef get_recommendations(user_id: str) -> list:\n \"\"\"Get personalized recommendations\"\"\"\n prefs = agent.vault.query('preferences', {'user_id': user_id})\n # Use preferences to customize recommendations\n return [\"item1\", \"item2\", \"item3\"]\n```\n\n## \ud83d\udd04 Multi-Provider Flexibility\n\nSwitch AI providers without changing code:\n\n```python\n# Use managed Bedrock (included in pricing)\nagent = Agent(\"my-agent\", model=\"claude-3-sonnet\")\n\n# Or bring your own OpenAI key \nagent = Agent(\"my-agent\", \n provider=\"openai\",\n model=\"gpt-4o\", \n byok={\"api_key\": \"sk-your-key\"}\n)\n\n# Same tools, same deployment, different AI provider\n```\n\n## \ud83d\udcca Real-World Examples\n\n### Slack Customer Support Bot\n```python\nagent = Agent(\"support-bot\")\n\n@agent.tool\ndef search_knowledge_base(query: str) -> str:\n # Search your docs/FAQ\n return search_results\n\n@agent.tool \ndef create_ticket(issue: str, priority: str) -> str:\n # Create Zendesk/Jira ticket\n return ticket_id\n\n# Deploy and connect to Slack webhook\nendpoint = agent.deploy()\n```\n\n### E-commerce Order Assistant\n```python\nagent = Agent(\"order-assistant\")\n\n@agent.tool\ndef check_inventory(product_id: str) -> dict:\n # Check your inventory system\n return {\"in_stock\": True, \"quantity\": 50}\n\n@agent.tool\ndef process_return(order_id: str, reason: str) -> str:\n # Handle return logic\n return \"Return processed\"\n\n# Deploy and embed in your website\nendpoint = agent.deploy()\n```\n\n### Content Creation Workflow \n```python\nagent = Agent(\"content-creator\")\n\n@agent.tool\ndef research_topic(topic: str) -> str:\n # Research from multiple sources\n return research_summary\n\n@agent.tool\ndef publish_to_cms(title: str, content: str) -> str:\n # Publish to WordPress/Contentful\n return \"Published successfully\"\n\n# Deploy and trigger via webhook\nendpoint = agent.deploy()\n```\n\n## \ud83d\udcb0 Simple Pricing\n\n**Session-based pricing.** One API call = one session, regardless of message count.\n\n- **Free**: 25 sessions/month\n- **Hobbyist**: $15/month, 200 sessions \n- **Starter**: $50/month, 1,000 sessions\n- **Professional**: $200/month, 5,000 sessions\n- **Enterprise**: Custom pricing, 25,000+ sessions\n\n## \ud83d\udcda Learn More\n\n- **\ud83d\udcd6 [Documentation](https://docs.flowstack.fun/)** - Complete guides and examples\n- **\ud83d\ude80 [5-Minute Tutorial](https://docs.flowstack.fun/quickstart/)** - Get started immediately\n- **\ud83e\udde0 [DataVault Guide](https://docs.flowstack.fun/datavault/)** - Persistent agent memory\n- **\ud83c\udf73 [Recipe Collection](https://docs.flowstack.fun/recipes/chatbot/)** - Real-world examples\n\n## \ud83d\udee0\ufe0f Development\n\n```bash\n# Local development\ngit clone https://github.com/flowstack-fun/flowstack.git\ncd flowstack\npip install -e .\n\n# Run tests \npytest tests/\n\n# Build docs\nmkdocs serve\n```\n\n## \ud83e\udd1d Support\n\n- **\ud83d\udc1b Issues**: [GitHub Issues](https://github.com/flowstack-fun/flowstack/issues)\n- **\ud83d\udcac Community**: [Discord](https://discord.gg/flowstack) \n- **\ud83d\udce7 Email**: [support@flowstack.fun](mailto:support@flowstack.fun)\n- **\ud83d\udc26 Updates**: [@flowstack](https://twitter.com/flowstack)\n\n---\n\n**Built for developers who ship fast.** \ud83d\udea2\n\n*Stop building infrastructure. Start building agents.*\n",
"bugtrack_url": null,
"license": null,
"summary": "AI Agent Platform SDK with built-in billing management",
"version": "1.0.0",
"project_urls": {
"Bug Reports": "https://github.com/flowstack-fun/flowstack/issues",
"Documentation": "https://docs.flowstack.fun",
"Homepage": "https://github.com/flowstack-fun/flowstack",
"Source": "https://github.com/flowstack-fun/flowstack"
},
"split_keywords": [
"ai",
" agents",
" llm",
" openai",
" anthropic",
" bedrock",
" aws"
],
"urls": [
{
"comment_text": null,
"digests": {
"blake2b_256": "7a98094364c94d82231c4d2855ee0c2fb209c92424e91b9dc7ad3b109f323b1d",
"md5": "d4b0abdb21e87d0c29487ae10ce5784b",
"sha256": "d58e0e3bbaaf388b59f15e32756f147a2acd7a519542e91a8c04c0898fcd5e53"
},
"downloads": -1,
"filename": "flowstack-1.0.0-py3-none-any.whl",
"has_sig": false,
"md5_digest": "d4b0abdb21e87d0c29487ae10ce5784b",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.8",
"size": 13023,
"upload_time": "2025-08-22T22:03:01",
"upload_time_iso_8601": "2025-08-22T22:03:01.236540Z",
"url": "https://files.pythonhosted.org/packages/7a/98/094364c94d82231c4d2855ee0c2fb209c92424e91b9dc7ad3b109f323b1d/flowstack-1.0.0-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": null,
"digests": {
"blake2b_256": "0dfe5137281f5bf3922030ce51c277dd77cf252b35f1473373aafc0a76cf78f2",
"md5": "c06b9888c7deada737c4d16d14bfd069",
"sha256": "1ab51edb820979688f33b4a43b2d3888dc115cced7da6ea2845359e2e7329f5b"
},
"downloads": -1,
"filename": "flowstack-1.0.0.tar.gz",
"has_sig": false,
"md5_digest": "c06b9888c7deada737c4d16d14bfd069",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.8",
"size": 14002,
"upload_time": "2025-08-22T22:03:02",
"upload_time_iso_8601": "2025-08-22T22:03:02.626221Z",
"url": "https://files.pythonhosted.org/packages/0d/fe/5137281f5bf3922030ce51c277dd77cf252b35f1473373aafc0a76cf78f2/flowstack-1.0.0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2025-08-22 22:03:02",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "flowstack-fun",
"github_project": "flowstack",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"lcname": "flowstack"
}
JSON
