glaip-sdk 0.0.5b1

Python SDK for GL AIP (GDP Labs AI Agent Package) - Simplified CLI Design

GL AIP SDK — Enterprise AI Agent Package (Python)

Build, run, and manage AI agents with a minimal, production-ready Python API.

• Simple: Client() → create resources → agent.run("...")
• Strong: typed models (Pydantic), robust errors, clean streaming renderer
• Enterprise-ready: shared HTTP session, timeouts/retries, consistent resource lifecycle
• Secure: automatic recursive secret masking; no telemetry; TLS recommended
• Modern CLI: rich terminal experience, subsequence-based fuzzy search, smart paging, multiple output formats

---

📋 Table of Contents

• Quick Start
• Features
• Latest Improvements
• Use Cases & Industries
• Performance & Scalability
• Core API
• CLI
• Configuration
• Learning Path
• Development
• Error Handling
• Troubleshooting
• Getting Help & Support
• Why Choose AIP SDK?

---

🚀 Installation

📋 Current Status: The AIP SDK is currently available for local development. Global installation from PyPI will be available once the package is published.

Option 1: Local Development Setup (Recommended for developers - Available Now)

# Clone the repository
git clone <repository-url>
cd ai-agent-platform-sdk

# Install in editable mode - CLI command 'aip' will be available globally
pip install -e .

# Verify installation
aip --help

Option 2: Poetry Development Setup (Alternative for Poetry users - Available Now)

# Clone the repository
git clone <repository-url>
cd ai-agent-platform-sdk

# Install dependencies and package using Poetry
poetry install

# Run CLI through Poetry (CLI not available globally)
poetry run aip --help

Option 3: Global Installation from PyPI (Coming Soon)

# Install globally - CLI command 'aip' will be available everywhere
# ⚠️  This option will be available once the package is published to PyPI
pip install glaip-sdk

# Verify installation
aip --help

Note: All required dependencies including click and rich are automatically installed with the package. This resolves the previous "click and rich not installed" errors that users encountered.

🛠️ Local Development Setup

Prerequisites

• Python 3.10+ installed on your system
• Git for cloning the repository
• pip or Poetry for package management

Quick Local Setup

# 1. Clone the repository
git clone <repository-url>
cd ai-agent-platform-sdk

# 2. Install in editable mode (recommended)
pip install -e .

# 3. Verify CLI is working
aip --version
aip --help

# 4. Test basic functionality
aip status

Development Workflow

# After making changes to the code:
pip install -e . --force-reinstall

# Or simply reinstall:
pip install -e .

# Test your changes:
aip --version

Testing Examples

# Run hello world examples
python3 examples/getting-started/sdk/01_hello_world_agent.py
python3 examples/getting-started/sdk/02_hello_world_tool.py

# List all available examples
python3 examples/run_examples.py --list

Updating Local Development

# Pull latest changes
git pull origin main

# Reinstall to get latest version
pip install -e . --force-reinstall

# Verify version
aip --version

# Test new features
aip --help

Troubleshooting Local Setup

Common Issues

"aip command not found" after pip install -e .

# Check if the package was installed correctly
pip list | grep glaip-sdk

# Reinstall with force flag
pip install -e . --force-reinstall

# Verify the CLI script was created
which aip

Permission errors during installation

# Use user installation (recommended)
pip install -e . --user

# Or use virtual environment
python3 -m venv venv
source venv/bin/activate
pip install -e .

Dependency conflicts

# Clean install
pip uninstall glaip-sdk -y
pip install -e . --no-deps
pip install -e .

⚡ Quick Start (30 seconds)

# Option 1: Local development setup (Available Now)
git clone <repository-url>
cd ai-agent-platform-sdk
pip install -e .

# Option 2: Install globally from PyPI (Coming Soon)
# pip install glaip-sdk

export AIP_API_KEY="your-api-key"
export AIP_API_URL="https://your-platform.com/api"

Python SDK:

from glaip_sdk import Client

client = Client()
agent = client.create_agent(name="hello", instruction="You are a helpful assistant.")
print(agent.run("Hello! What can you do?"))
agent.delete()

CLI:

# Configure and create your first agent
aip configure
aip agents create --name "hello" --instruction "You are a helpful assistant"
aip agents run <AGENT_ID> --input "Hello! What can you do?"
aip agents delete <AGENT_ID>

That's it. You created, ran, and cleaned up your first agent. 🎉

---

🔧 Connection & Timeouts

from glaip_sdk import Client

# Auto-config from environment
with Client(timeout=60.0) as client:   # shared HTTP session, closes on exit
    print(len(client.list_agents()))

# Or explicit configuration
client = Client(
    api_url="https://your-platform.com/api",
    api_key="your-key",
    timeout=120.0
)

---

✨ Features

• Python SDK: Simple Client() → create resources → agent.run("...")
• CLI Interface: Full command-line control with aip commands
• Typed Models: Pydantic-powered Agent, Tool, MCP with .run(), .update(), and .delete() methods
• Streaming UX: Live markdown, tool/sub-agent steps, and theme options (AIP_THEME, AIP_NO_EMOJI)
• Shared Session: All sub-clients share one HTTP session for reliability and performance
• 🔐 Enterprise Security: Automatic recursive secret masking for sensitive data
• 📱 Multiple Output Formats: Rich, JSON, Plain text, and Markdown views
• 🎯 Modern CLI Experience: Subsequence-based fuzzy search, smart paging, and intuitive navigation
• ⚡ Performance: Optimized HTTP sessions, connection pooling, and efficient streaming
• 🛡️ Reliability: Comprehensive error handling, configurable timeouts, and graceful degradation
• 🔧 Developer Tools: Rich debugging, comprehensive logging, and testing utilities

---

🎯 Use Cases & Industries

🚀 Development & DevOps

• CI/CD Automation: Deploy agents to manage build pipelines and deployments
• Code Review: Automated code analysis and quality checks
• Infrastructure Management: Cloud resource provisioning and monitoring
• Testing Automation: Intelligent test generation and execution

💼 Business Operations

• Customer Support: AI-powered customer service agents
• Data Analysis: Automated insights and reporting
• Process Automation: Streamline repetitive business workflows
• Document Processing: Intelligent document analysis and summarization

🔬 Research & Analytics

• Data Processing: Automated data cleaning and transformation
• Research Automation: Literature review and data synthesis
• Model Training: Automated ML pipeline management
• Reporting: Dynamic report generation and insights

🏢 Enterprise Applications

• Compliance: Automated regulatory compliance checking
• Security: Threat detection and incident response
• Auditing: Automated audit trail analysis
• Integration: Connect with existing enterprise systems

---

🚀 Performance & Scalability

⚡ High Performance

• Connection Pooling: Efficient HTTP session reuse across requests
• Streaming Responses: Real-time data processing without memory bloat
• Optimized Serialization: Fast JSON parsing and response handling
• Concurrent Operations: Support for parallel agent execution

📈 Scalability Features

• Resource Management: Automatic cleanup and lifecycle management
• Memory Efficiency: Streaming processing for large datasets
• Timeout Handling: Configurable timeouts for long-running operations
• Connection Management: Efficient HTTP session reuse and connection pooling

🏗️ Enterprise Architecture

• Microservices Ready: Designed for distributed system integration
• API-First Design: RESTful endpoints with consistent patterns
• Monitoring Support: Built-in logging and debugging capabilities
• Security Compliance: Enterprise-grade security and audit features

Security & Data Handling:

• 🔐 Secret Masking: Sensitive data automatically masked in JSON, Plain, and Markdown outputs
• 🔑 API Key Security: API key read from env/config; never logged or persisted by default
• 📊 No Telemetry: Zero data collection or external reporting
• 💡 Best Practice: Use export AIP_API_KEY="..." instead of command-line flags

Windows Compatibility:

• 📄 Pager Behavior: On Windows, automatically uses Rich's built-in pager for better ANSI support

Interactive Features (Optional Dependencies):

• 🔍 Fuzzy Search: prompt_toolkit for intelligent subsequence-based search
• 📋 Select Menus: questionary for interactive selection menus
• 🔄 Fallback Behavior: Falls back to Rich tables when fuzzy search unavailable or non-TTY

---

🆕 Latest Improvements

🔐 Enhanced Security with Recursive Secret Masking

Automatically masks sensitive fields like API keys, tokens, and secrets at all levels of nested data structures:

# Sensitive data is automatically masked in all views
aip agents get <AGENT_ID>  # API keys shown as "••••••••"
aip agents list --view json  # Recursive masking in JSON output

📱 Multiple Output Formats

Choose your preferred output format for any command:

# Rich terminal experience (default)
aip agents list

# Clean JSON for scripting
aip agents list --view json

# Simple plain text
aip agents list --view plain

# Markdown tables
aip agents list --view md

🎯 Enhanced CLI Experience

• Fuzzy Search: Intelligent search with typo tolerance
• Smart Paging: Automatic paging for large datasets
• Rich Rendering: Beautiful tables and panels with proper TTY detection

---

Core API in 60 seconds

Create & run an agent

from glaip_sdk import Client

client = Client()
agent = client.create_agent(
    name="helper",
    instruction="You are a helpful assistant."
)
print(agent.run("Summarize why streaming is useful."))

Add a tool, then use it

tool = client.create_tool(
    file_path="my_tool.py",
    name="my_tool",          # Optional: extracted from file if not provided
    description="Custom business logic"  # Optional: extracted from file if not provided
)

agent = client.create_agent(
    name="tool-user",
    instruction="Use tools when helpful.",
    tools=[tool.id]
)

print(agent.run("Use the custom tool to process today's data."))

Discover models

for lm in client.list_language_models():
    print(lm["name"], lm["provider"])

Update & delete

agent.update(instruction="Be concise.")
agent.delete()
tool.delete()

---

CLI

# Configure
aip configure

# Agents
aip agents create --name "helper" --instruction "You are helpful"
aip agents run <AGENT_ID> --input "Hello!"
aip agents delete <AGENT_ID>

# Tools
aip tools create --file my_tool.py --name my_tool --description "Custom logic"
aip tools list

# Multiple output formats
aip agents list --view json      # JSON for scripting
aip agents list --view plain     # Simple text
aip agents list --view md        # Markdown tables

Run full examples: python -m examples Getting started: examples/getting-started/sdk/01_hello_world_agent.py

---

Configuration

• Environment variables

  # Copy the example file and fill in your values
  cp .env.example .env
  
  # Or set manually
  export AIP_API_KEY="your-api-key"
  export AIP_API_URL="https://your-platform.com/api"
  

• Optional config file: ~/.aip/config.yaml

  api_url: https://your-platform.com/api
  api_key: your-api-key
  

Configuration precedence: CLI context flags ⟶ environment variables ⟶ ~/.aip/config.yaml

• Renderer options

  export AIP_THEME=dark        # or "light"
  export AIP_NO_EMOJI=true     # disable emojis in streaming view
  export AIP_PAGER_HEADER=0    # disable pager headers
  

• Security options

  export AIP_MASK_OFF=1        # disable secret masking (not recommended)
  export AIP_MASK_FIELDS="custom_field,another_secret"  # add custom fields to mask
  

Environment Variables Reference:

Variable	Default	Purpose
AIP_API_URL	—	Base API endpoint
AIP_API_KEY	—	API key used for auth (never logged)
AIP_TIMEOUT	30.0	Request timeout (s)
AIP_THEME	dark	Streaming code theme (dark/light)
AIP_NO_EMOJI	false	Disable emojis in the stream view
AIP_PAGER_HEADER	1	Show pager header (0/1)
AIP_PAGER_WRAP	0	Wrap lines in pager (drops -S)
AIP_LESS_FLAGS	auto	Override default less flags
AIP_MASK_OFF	0	Disable secret masking (not recommended)
AIP_MASK_FIELDS	—	Extra comma-separated keys to mask
AIP_TABLE_NO_SORT	0	Disable automatic table sorting

• Enterprise features

  • Single shared HTTP session across sub-clients for reliability & connection reuse
  • Configurable timeouts and SSE streaming out of the box
  • Proxy support: HTTP_PROXY / HTTPS_PROXY supported by httpx
  • Python: 3.10–3.12
  • Versioning: semantic versioning; breaking changes bump MAJOR
  • Security: API key only read from env/config, never logged; no telemetry by default; TLS recommended

---

Learning Path

• Start here

  • examples/getting-started/sdk/01_hello_world_agent.py
  • examples/getting-started/sdk/02_hello_world_tool.py
  • examples/getting-started/sdk/03_hello_world_mcp.py

• Build skills

  • examples/intermediate/sdk/01_conversation_memory.py
  • examples/intermediate/sdk/02_streaming_execution.py
  • examples/intermediate/sdk/04_agent_tool_integration.py

• Go deeper

  • examples/advanced/sdk/01_agent_workflow_orchestration.py
  • examples/advanced/sdk/02_error_handling_patterns.py
  • examples/advanced/sdk/03_performance_optimization.py
  • examples/advanced/sdk/04_metadata_patterns.py

• SDK Testing Scenario See: SDK_TESTING_SCENARIO.md (complete end-to-end coverage plan)

---

Development

make install
make test-unit
make test-integration
make pre-commit

Versioning

Single source of truth: pyproject.toml under [project].version.

The SDK resolves its runtime version from installed package metadata (glaip_sdk/_version.py), so you do not need to edit any code files when changing the version. Update once in pyproject.toml, then build/reinstall.

Release by tagging (e.g., v0.0.3) and creating a GitHub Release that matches the tag.

---

🚨 Error Handling

from glaip_sdk import Client
from glaip_sdk.exceptions import (
    AuthenticationError, NotFoundError, ValidationError,
    RateLimitError, TimeoutError, ServerError, AIPError
)

try:
    client = Client()
    agent = client.create_agent(name="demo", instruction="...")
    print(agent.run("hello"))
except AuthenticationError:
    print("Invalid API key or insufficient permissions.")
except NotFoundError:
    print("Resource not found.")
except RateLimitError:
    print("Rate limited — retry later.")
except TimeoutError:
    print("Request timed out.")
except ValidationError as e:
    print(f"Bad request: {e}")
except ServerError as e:
    print(f"Server issue: {e}")
except AIPError as e:
    print(f"SDK error: {e}")

---

🔍 Troubleshooting

# Test the connection (auth header is required)
curl -s -H "X-API-Key: $AIP_API_KEY" "$AIP_API_URL/health-check"

# Verify your configuration
echo "API URL: $AIP_API_URL"
echo "API Key length: ${#AIP_API_KEY}"

# Test CLI functionality
aip --version
aip agents list --view json

---

🆘 Getting Help & Support

📚 Documentation & Examples

• Comprehensive Examples: Run python -m examples for complete working samples
• Getting Started: Start with examples/getting-started/ for basic concepts
• Advanced Patterns: Explore examples/advanced/ for complex use cases
• Testing Guide: See SDK_TESTING_SCENARIO.md for testing strategies

🔧 Troubleshooting

• Connection Issues: Check API URL and authentication
• Performance: Monitor timeouts and connection pooling
• Security: Verify secret masking and access controls
• CLI Issues: Test with --view json for debugging

💡 Best Practices

• Resource Management: Always clean up resources with .delete()
• Error Handling: Use try-catch blocks for robust applications
• Configuration: Use environment variables for sensitive data
• Testing: Test with small datasets before scaling up

---

🏆 Why Choose AIP SDK?

• 🚀 Production Ready: Built for enterprise with proper error handling, timeouts, and security
• 🔐 Security First: Automatic secret masking, no sensitive data logging
• 💻 Developer Experience: Intuitive CLI with fuzzy search and multiple output formats
• 📊 Rich Streaming: Beautiful terminal experience with live updates and tool visualization
• 🔧 Simple API: Minimal boilerplate, maximum functionality
• 🏗️ Enterprise Features: Shared sessions, proxy support, comprehensive error handling

---

License

MIT — see LICENSE