Skip to main content

AI Agent session management SDK

Project description

AgentAuth

AgentAuth

Persistent auth for AI agents. Works on any site, today.

No website integration required. No OAuth adoption needed. Just secure session management that works.

License: MIT PyPI version Python 3.8+ Dev.to

Quick StartLangChainn8nHow It WorksRoadmap


The Problem

# What every agent developer does today
cookies = {"session_id": "abc123..."}  # Hardcoded. Breaks constantly. Security nightmare.

Sessions expire. Cookies leak into git history. No audit trail. No access control.

The Solution

from agent_auth import Vault

vault = Vault()
cookies = vault.get_session("linkedin.com")  # Encrypted. Scoped. Audited.
pip install agentauth-py

Works on sites as they exist today. No website changes. No OAuth adoption. No waiting for the ecosystem.


See It In Action

One command to grab cookies from Chrome

Agent automatically authenticates using stored session—no manual login, no hardcoded cookies.


How It Works

  1. Log into any site in Chrome (you probably already are)
  2. Run agent-auth grab <domain> — cookies are encrypted and stored locally
  3. Your agent retrieves them securely on-demand

No extension required. No copy-paste. One command.


Quick Start

pip install agentauth-py
agent-auth grab linkedin.com

That's it. Your LinkedIn session is now stored in an encrypted vault.

The grab command reads cookies directly from Chrome—no extension needed, no manual export.

from agent_auth import Vault

vault = Vault()
cookies = vault.get_session("linkedin.com")

# Use with requests, Playwright, or any HTTP library
import requests
response = requests.get("https://linkedin.com/feed", cookies=cookies)

Works with any site you're logged into:

agent-auth grab github.com
agent-auth grab notion.so
agent-auth grab twitter.com

Why AgentAuth?

Without AgentAuth With AgentAuth
Hardcoded cookies in code Encrypted vault storage
Sessions in git history Secrets separate from code
Any code can access anything Scoped access per agent
No idea what accessed what Full audit logging
Sessions break silently Expiration notifications

LangChain Integration

from langchain_openai import ChatOpenAI
from langchain.agents import initialize_agent, AgentType
from agent_auth.langchain import get_agentauth_tools

tools = get_agentauth_tools(
    agent_name="sales-bot",
    vault_password="your-vault-password"  # Password auto-retrieved from system keyring if not provided
)

llm = ChatOpenAI(model="gpt-4", temperature=0)
agent = initialize_agent(
    tools=tools,
    llm=llm,
    agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION,
    verbose=True
)

# Agent automatically uses stored sessions
response = agent.run("Get my LinkedIn notifications")

Available tools:

  • authenticated_request — Make HTTP requests with stored session cookies
  • get_session_cookies — Retrieve cookies for custom requests

n8n Integration

Use AgentAuth in your n8n workflows with our community node.

npm install n8n-nodes-agentauth

The node lets you:

  • Store and retrieve session cookies in n8n workflows
  • Use authenticated sessions with HTTP Request nodes
  • Build automation flows that require login

📦 View on npm | GitHub


Playwright Integration

from playwright.sync_api import sync_playwright
from agent_auth import Vault

vault = Vault()
cookies = vault.get_session("github.com")

with sync_playwright() as p:
    browser = p.chromium.launch()
    context = browser.new_context()
    context.add_cookies(cookies)
    
    page = context.new_page()
    page.goto("https://github.com/notifications")
    # Already authenticated!

CLI Reference

Command Description
agent-auth init Initialize encrypted vault
agent-auth grab <domain> Grab cookies directly from Chrome
agent-auth add <domain> Add session for a domain
agent-auth list List stored sessions
agent-auth create-agent <n> --scopes <domains> Create scoped agent
agent-auth agents List registered agents

Security

  • AES-256 encryption with PBKDF2 key derivation (100k iterations)
  • Ed25519 keypairs for agent identity
  • Scoped access — agents only access approved domains
  • Audit logging — every access logged with timestamp
  • SQLite storage — encrypted database at ~/.agent-auth/vault.db

Roadmap

  • Chrome extension for cookie export
  • Encrypted local vault
  • Direct Chrome cookie extraction (agent-auth grab)
  • Playwright integration
  • LangChain integration
  • n8n integrationn8n-nodes-agentauth
  • Firefox extension
  • Selenium examples
  • Cloud vault sync

Examples

Working demos available in the examples/ folder.

GitHub Agent Demo

Setup:

# 1. Create the agent
agent-auth create-agent github-agent --scopes github.com

# 2. Export cookies from github.com using Chrome extension

# 3. Add the session
agent-auth add github.com

Run:

PYTHONPATH=. python examples/github_agent.py

Contributing

PRs welcome. Check out the issues for feature requests.


License

MIT

Star us if AgentAuth helps your agents authenticate!

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

agentauth_py-0.2.2.tar.gz (21.0 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

agentauth_py-0.2.2-py3-none-any.whl (22.7 kB view details)

Uploaded Python 3

File details

Details for the file agentauth_py-0.2.2.tar.gz.

File metadata

  • Download URL: agentauth_py-0.2.2.tar.gz
  • Upload date:
  • Size: 21.0 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.9.6

File hashes

Hashes for agentauth_py-0.2.2.tar.gz
Algorithm Hash digest
SHA256 0ca9f0602c4cd244ce6f9ff973ba24a2e217ce011f3c3b6edc86f1fc9665a754
MD5 d163e124cd5965e130d4a06f96624bf5
BLAKE2b-256 75fd997aeb58d2d7b54132d83e0b717c9954045bf15274eaa515cf9de818f04e

See more details on using hashes here.

File details

Details for the file agentauth_py-0.2.2-py3-none-any.whl.

File metadata

  • Download URL: agentauth_py-0.2.2-py3-none-any.whl
  • Upload date:
  • Size: 22.7 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.9.6

File hashes

Hashes for agentauth_py-0.2.2-py3-none-any.whl
Algorithm Hash digest
SHA256 8a0836d5f930a2e94628bb09596c71252fbb9e71a59c3c2521d7d7b467e0c19a
MD5 b0448dd157ec2e39145fc0db4a2eb08c
BLAKE2b-256 f9cf98182467235442b0a5ef0530460fa6c7d0980393007756c0a850eb92bc91

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page