fibonacci-sdk 0.2.4

Python SDK for Fibonacci Workflow Automation Platform

Fibonacci SDK

Build and deploy AI-powered workflows programmatically

Documentation • Installation • Quick Start • Features • Issues

---

What is Fibonacci?

Fibonacci is a platform for building autonomous AI agents and workflows. The SDK allows developers to:

• 🤖 Build AI Workflows - Chain LLM calls, tools, and logic programmatically
• 🚀 Deploy Instantly - One command to deploy to production
• 🔧 40+ Integrations - Connect to Slack, Google Sheets, GitHub, and more
• 📊 Monitor & Optimize - Track costs, performance, and optimize automatically

---

Installation

pip install fibonacci-sdk

With Security Features (Recommended)

pip install fibonacci-sdk[security]

This includes secure keychain storage for API keys.

---

Quick Start

1. Get Your API Key

Sign up at fibonacci.today to get your API key.

2. Set Up Authentication

# Option 1: Environment variable
export FIBONACCI_API_KEY="fib_live_your_api_key"

# Option 2: Secure keychain storage (recommended)
fibonacci security save
# Then enter your API key when prompted

3. Create Your First Workflow

from fibonacci import Workflow, LLMNode, ToolNode

# Create a workflow
wf = Workflow(
    name="Customer Support Bot",
    description="AI-powered customer support assistant"
)

# Add an LLM node
respond = LLMNode(
    id="respond",
    name="Generate Response",
    instruction="""
    You are a helpful customer support assistant.
    
    Customer message: {{input.message}}
    
    Provide a friendly, helpful response.
    """
)

wf.add_node(respond)

# Deploy to Fibonacci platform
workflow_id = wf.deploy()
print(f"Deployed! Workflow ID: {workflow_id}")

# Execute the workflow
result = wf.run(input_data={"message": "How do I reset my password?"})
print(result.output_data)

---

Features

🔗 Node Types

Node	Description	Example Use Case
LLMNode	Call Claude AI models	Text generation, analysis, summarization
ToolNode	Execute platform tools	Read Google Sheets, send Slack messages
CriticNode	Evaluate outputs	Quality scoring, validation
ConditionalNode	Branch logic	Route based on sentiment, conditions

📦 Example: Multi-Step Workflow

from fibonacci import Workflow, LLMNode, ToolNode, ConditionalNode

wf = Workflow(name="Sales Report Pipeline")

# Step 1: Read data from Google Sheets
read_data = ToolNode(
    id="read_data",
    name="Read Sales Data",
    tool="google_sheets_read",
    params={"spreadsheet_id": "{{input.sheet_id}}"}
)

# Step 2: Analyze with AI
analyze = LLMNode(
    id="analyze",
    name="Analyze Sales",
    instruction="Analyze this sales data and identify trends: {{read_data}}",
    dependencies=["read_data"]
)

# Step 3: Check if urgent
check_urgent = ConditionalNode(
    id="check_urgent",
    name="Check Urgency",
    left_value="{{analyze}}",
    operator="contains",
    right_value="urgent",
    true_branch=["send_alert"],
    false_branch=["send_report"],
    dependencies=["analyze"]
)

# Step 4a: Send alert if urgent
send_alert = ToolNode(
    id="send_alert",
    name="Send Urgent Alert",
    tool="slack_send_message",
    params={
        "channel": "#sales-alerts",
        "message": "🚨 URGENT: {{analyze}}"
    },
    dependencies=["check_urgent"]
)

# Step 4b: Send regular report
send_report = ToolNode(
    id="send_report",
    name="Send Report",
    tool="slack_send_message",
    params={
        "channel": "#sales-reports",
        "message": "📊 Daily Report: {{analyze}}"
    },
    dependencies=["check_urgent"]
)

wf.add_nodes([read_data, analyze, check_urgent, send_alert, send_report])

# Deploy and run
wf.deploy()
result = wf.run(input_data={"sheet_id": "abc123"})

💾 Memory & State

Persist data across workflow runs:

from fibonacci import Memory

# Store data
memory = Memory(scope="workflow", workflow_id="your-workflow-id")
memory.set("user_preferences", {"theme": "dark", "language": "en"})

# Retrieve later
prefs = memory.get("user_preferences")
print(prefs)  # {"theme": "dark", "language": "en"}

📄 YAML Support

Define workflows in YAML:

# workflow.yaml
name: Email Summarizer
description: Summarize incoming emails

nodes:
  - id: summarize
    type: llm
    name: Summarize Email
    instruction: "Summarize this email: {{input.email_body}}"
    config:
      model: claude-haiku-4-5
      max_tokens: 500

from fibonacci import Workflow

# Load from YAML
wf = Workflow.from_yaml("workflow.yaml")
wf.deploy()

# Export to YAML
wf.to_yaml("exported.yaml")

---

CLI Commands

# Initialize a new workflow project
fibonacci init "My Workflow"

# List your workflows
fibonacci list

# Execute a workflow
fibonacci run <workflow-id> '{"message": "hello"}'

# Check run status
fibonacci status <run-id>

# Security commands
fibonacci security status    # Check API key security
fibonacci security save      # Save API key to secure keychain
fibonacci security migrate   # Migrate from .env to keychain

# Audit logs
fibonacci audit view         # View recent audit events

---

Configuration

Environment Variables

Variable	Description	Default
FIBONACCI_API_KEY	Your API key	Required
FIBONACCI_BASE_URL	API base URL	http://api.fibonacci.today
FIBONACCI_TIMEOUT	Request timeout (seconds)	300
FIBONACCI_DEBUG	Enable debug logging	false

Configuration File

Create ~/.fibonacci/config.yaml:

api_key: fib_live_your_key
base_url: http://api.fibonacci.today
timeout: 300
debug: false

Secure Storage (Recommended)

Store your API key in the system keychain:

fibonacci security save

This uses:

• macOS: Keychain
• Windows: Credential Manager
• Linux: Secret Service

---

Tool Discovery

Find available tools programmatically:

from fibonacci import list_tools, get_tool_schema, search_tools

# List all tools
tools = list_tools()
for tool in tools:
    print(f"{tool['name']}: {tool['description']}")

# Get detailed schema
schema = get_tool_schema("google_sheets_read")
print(schema)

# Search for tools
results = search_tools("slack")
print(results)

---

Error Handling

from fibonacci import (
    Workflow,
    FibonacciError,
    AuthenticationError,
    ValidationError,
    ExecutionError
)

try:
    wf = Workflow(name="My Workflow")
    wf.deploy()
except AuthenticationError:
    print("Invalid API key")
except ValidationError as e:
    print(f"Workflow validation failed: {e.errors}")
except ExecutionError as e:
    print(f"Execution failed: {e.message}")
except FibonacciError as e:
    print(f"General error: {e.message}")

---

Requirements

• Python 3.11 or higher
• Valid Fibonacci API key

Dependencies

• httpx - HTTP client
• pydantic - Data validation
• typer - CLI framework
• rich - Terminal formatting
• pyyaml - YAML support
• keyring (optional) - Secure credential storage

---

Documentation

• 📖 Full Documentation: docs.fibonacci.today
• 🔧 API Reference: docs.fibonacci.today/api

---

License

This SDK is proprietary software. See LICENSE for terms.

TL;DR: You can use this SDK with a valid Fibonacci API key. You cannot modify, redistribute, or create competing products.

---

Contributing

We welcome bug reports and feature suggestions! See CONTRIBUTING.md for guidelines.

---

Built with ❤️ by Fibonacci, Inc.

Website • Docs •