Skip to main content

A sophisticated Security Liaison and Governance Layer for AI Agents

Project description

🛡️ AegisFlow

The Universal Security Layer for AI Agents.

AegisFlow is a CPU-efficient Python library that sits between your LLM and your system. It prevents prompt injections, scrubs sensitive data (API keys/PII), and enforces a "Human-in-the-Loop" protocol for high-risk operations.

AegisFlow Dashboard

✨ Features

  • Sentinel State Engine: Tracks agent behavior over time and escalates threats based on "risk streaks."
  • Behavioral Redlines: Detects recursive deletions (rm -rf), unauthorized POST requests, and security bypass attempts.
  • Key Scrubber: Automatically redacts API keys and EMV data before they leak to the LLM.
  • Human-in-the-Loop: Requires a "Reasoning String" justification for any high-risk system commands.

🚀 Quick Start

pip install aegisflow

Scan your agent scripts for vulnerabilities:

aegis scan my_agent_script.py

🏗️ Architecture

AegisFlow operates as a lightweight governance layer. It intercepts function calls and network requests, assigning a Threat Level (Low, Medium, High) to each action.

  • Low Risk: Allowed and logged.
  • Medium Risk: Warned and logged; contributes to a "Risk Streak."
  • High Risk: Blocked unless the user provides a valid Reasoning String.

📊 Sentinel Reports

View your security audit logs in a professional terminal dashboard:

aegis report

Example Output:

+-----------------------------------------------------------------------------+
| Timestamp                | Level  | Action      | Outcome     | Details     |
|--------------------------+--------+-------------+-------------+-------------|
| 2026-02-10T12:34:07-0500 | Low    | safe_op     | EXECUTED    | {'content': |
| 2026-02-10T23:03:31-0500 | High   | file_op     | USER_OVERR | {'path':    |
|                          |        |             |             | '/etc/shad |
+-----------------------------------------------------------------------------+

📦 Installation & Setup

  1. Install via Pip:
    pip install aegisflow
    
  2. Initialize Configuration (Optional): Create a .aegis.json in your project root:
    {
      "protected_paths": ["/prod/db", "./secrets"],
      "strict_mode": true
    }
    

🤝 Universal AI Integration

AegisFlow is designed to wrap any LLM provider. (Coming soon: Native LiteLLM integration).

License

MIT

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

aegisflow-2.0.0.tar.gz (9.3 kB view details)

Uploaded Source

Built Distribution

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

aegisflow-2.0.0-py3-none-any.whl (12.1 kB view details)

Uploaded Python 3

File details

Details for the file aegisflow-2.0.0.tar.gz.

File metadata

  • Download URL: aegisflow-2.0.0.tar.gz
  • Upload date:
  • Size: 9.3 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.2

File hashes

Hashes for aegisflow-2.0.0.tar.gz
Algorithm Hash digest
SHA256 3d0888822fcb214ba3be2ad6b5ab34e40c191885a5aba215beb5d6c823a7d399
MD5 f8180bedec211329017e77092f30eedb
BLAKE2b-256 37bcc17ecd63f47cb811599536825d56ede1da4ec2dffa807ba3e95515bb196f

See more details on using hashes here.

File details

Details for the file aegisflow-2.0.0-py3-none-any.whl.

File metadata

  • Download URL: aegisflow-2.0.0-py3-none-any.whl
  • Upload date:
  • Size: 12.1 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.2

File hashes

Hashes for aegisflow-2.0.0-py3-none-any.whl
Algorithm Hash digest
SHA256 edbb6b22dd3dd74d8517d59a31d508deacbbb6387018c932c799842f1b7d3126
MD5 413d3c45122d4aadbee831ad486e198a
BLAKE2b-256 a2f9089995aa1015592df08ac28929c874362154cb7809b0166157b92d2c2420

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