AgentGuard Python SDK
The official Python SDK for AgentGuard Pro - Enterprise AI governance and compliance platform.
🚀 One-Line Integration
Transform your AI infrastructure with a single line of code:
from agentguard import AgentGuard
# Initialize once
agentguard = AgentGuard(tenant_id="your-tenant")
# Before: Complex manual compliance
response = openai.ChatCompletion.create(model="gpt-4", messages=[...])
# After: Automatic compliance with intercept()
response = agentguard.intercept(
openai.ChatCompletion.create,
model="gpt-4",
messages=[...]
)
That's it! The intercept() method automatically handles:
- ✅ Authorization checks
- ✅ Provider detection
- ✅ Context extraction
- ✅ Audit logging
- ✅ Error handling
Installation
pip install agentguard
Quick Start
1. Set up your environment
export AGENTGUARD_API_KEY="your-api-key"
export AGENTGUARD_TENANT_ID="your-tenant-id"
2. Use with any AI provider
from agentguard import AgentGuard
import openai
# Initialize AgentGuard
agentguard = AgentGuard()
# Intercept any AI call
response = agentguard.intercept(
openai.ChatCompletion.create,
model="gpt-4",
messages=[
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "Hello!"}
]
)
Supported Providers
The SDK automatically detects and supports:
- ✅ OpenAI (GPT-4, GPT-3.5, DALL-E, etc.)
- ✅ Anthropic (Claude)
- ✅ Azure OpenAI
- ✅ Custom providers (via registration)
Advanced Usage
Custom Provider Registration
# Register your custom AI provider
agentguard.register_provider(
name='custom-llm',
detect_fn=lambda fn, args, kwargs: 'custom' in str(fn),
extract_model_fn=lambda fn, args, kwargs: kwargs.get('model', 'custom'),
extract_context_fn=lambda fn, args, kwargs: {'custom': True},
resource='ai:custom'
)
# Now intercept() works with your custom provider!
response = agentguard.intercept(custom_llm.generate, prompt="Hello")
Error Handling
from agentguard import AgentGuardError
try:
response = agentguard.intercept(
openai.ChatCompletion.create,
model="gpt-4",
messages=[{"role": "user", "content": "Sensitive request"}]
)
except AgentGuardError as e:
if e.code == 'AGENTGUARD_ACCESS_DENIED':
print(f"Access denied: {e}")
# Handle authorization failure
Direct Authorization (Advanced)
# For fine-grained control, use authorize() directly
auth_result = agentguard.authorize(
subject="user@example.com",
resource="ai:gpt-4",
action="generate",
context={"purpose": "customer_support"}
)
if auth_result['allowed']:
# Proceed with AI call
pass
Legacy Context Manager (Deprecated)
The SDK still supports the legacy context manager pattern:
from agentguard import protect
with protect(context={"user_id": "123"}) as guard:
response = openai.ChatCompletion.create(...)
guard.complete(response)
Configuration
Environment Variables
AGENTGUARD_API_KEY: Your API key (required)AGENTGUARD_TENANT_ID: Default tenant IDAGENTGUARD_PRINCIPAL: Default principal/userAGENTGUARD_BASE_URL: API base URL (defaults to https://api.agentguard.pro)
Initialization Options
agentguard = AgentGuard(
api_key="your-api-key", # Or use env var
tenant_id="your-tenant", # Or use env var
principal="user@example.com", # Or use env var
base_url="https://custom.api" # Or use env var
)
Examples
See the examples/ directory for complete examples:
- intercept_demo.py - Basic intercept() usage
- openai_integration.py - OpenAI integration
- More examples coming soon!
Support
- 📧 Email: support@agentguard.pro
- 📚 Documentation: https://docs.agentguardpro.com
- 💬 Discord: https://discord.gg/agentguard
Metadata
Release files for agentguard-pro 0.1.4
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| agentguard_pro-0.1.4.tar.gz | 10.7 kB | Without provenance | Details for agentguard_pro-0.1.4.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| agentguard_pro-0.1.4-py3-none-any.whl | Python 3 | none | any | Details for agentguard_pro-0.1.4-py3-none-any.whl |
Total release size: 21.4 kB
Release files / agentguard_pro-0.1.4.tar.gz
| Download URL | agentguard_pro-0.1.4.tar.gz |
|---|---|
| Size | 10.7 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
b78c7161a9f4d5003060996249da8375b8b2076ff440ee37cb9eae3dbb5aba94
|
|
BLAKE2b-256 checksum How to use checksums |
f28e07cf26e567ac86172637e7849a16610ca803a9bc4ba59c97ba4312ae5b97
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.13.3
|
Release files / agentguard_pro-0.1.4-py3-none-any.whl
| Download URL | agentguard_pro-0.1.4-py3-none-any.whl |
|---|---|
| Size | 10.7 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
296236edfc4ee5f55fd2e8695422ed35b955886021f8bd902ce46c8f09008069
|
|
BLAKE2b-256 checksum How to use checksums |
aed7c088400e53890fb8866a702430ac9dd8bba5d3b710972315e21ac12e5eb5
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.13.3
|
Security for agentguard-pro 0.1.4
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.