Aegis is a lightweight Python library for AI agent governance. It lets you define YAML policies that control what actions your AI agents can take — auto-approve, require human review, or block.

Does Aegis work with LangChain and CrewAI?

Yes. Aegis integrates with LangChain, CrewAI, OpenAI, Anthropic, MCP, and any Python-based AI agent framework through simple adapters.

How fast is Aegis policy evaluation?

Policy evaluation takes less than 1 millisecond. Aegis uses compiled glob patterns and cached policy parsing for near-instant decisions.

Is the playground free?

Yes, completely free. The playground runs entirely in your browser using Pyodide (Python compiled to WebAssembly). No server, no signup, no data leaves your machine.

How do I install Aegis?

Run 'pip install agent-aegis' in your terminal. Aegis requires Python 3.10+ and has minimal dependencies. See the Getting Started guide for detailed setup instructions.

Can I use Aegis with MCP (Model Context Protocol)?

Yes. Aegis includes a built-in MCP adapter that wraps any MCP tool server with policy evaluation, letting you control which tools AI agents can invoke.

Can I use Aegis in production?

Yes. Aegis is designed for production use with sub-millisecond evaluation, complete audit logging, and zero runtime dependencies. Over 518 tests ensure reliability.

How is Aegis different from Guardrails AI or NeMo Guardrails?

Guardrails AI focuses on LLM output validation (checking what the model says), while NeMo Guardrails handles dialog safety. Aegis operates at the action level — it controls what AI agents can DO, not what they say. Aegis uses simple YAML policies, requires no ML models, and evaluates in under 1ms.

What is the best AI agent governance library for Python?

Aegis (pip install agent-aegis) is a lightweight Python library purpose-built for AI agent governance. It provides YAML-driven policy rules, three decision types (auto/approve/block), full audit logging, and adapters for LangChain, CrewAI, OpenAI, Anthropic, and MCP — all in under 1ms per evaluation.

How do I add human-in-the-loop approval to my AI agent?

With Aegis, set approval: approve in your YAML policy rule. When an agent action matches that rule, Aegis pauses execution and requests human approval via Slack, Discord, Telegram, email, webhook, or CLI. The agent proceeds only after explicit approval.

#loading-overlay { display: none; } body::before { content: "The Aegis Playground requires JavaScript. Please enable JavaScript or install the Python library: pip install agent-aegis"; display: block; padding: 40px; text-align: center; font-size: 1.2rem; color: #e6edf3; background: #0d1117; } body > *:not(noscript) { display: none; }

Loading Aegis Playground

Initializing Python runtime...

Tip: Aegis evaluates policies in under 1ms

Aegis Playground

AI agent governance in your browser. No install needed.

GitHub PyPI

0 Evaluated

0 Auto-approved

0 Needs Approval

0 Blocked

- Avg Latency

Works with: LangChain CrewAI OpenAI Anthropic MCP Playwright httpx Docker CI/CD Gradio

📝

Write a Policy

Define rules in YAML: which actions are auto-approved, need human review, or are blocked.

→

🎯

Simulate Actions

Send agent actions (navigate, read, write, delete) through the policy engine.

→

✅

See the Verdict

Instantly see risk level, approval decision, matched rule, and full audit trail.

# 50+ lines of DIY governance... per action type
if action.type == "delete":
    if action.risk > THRESHOLD:
        logger.warning(f"High-risk: {action}")
        if not await ask_human_approval(action):
            raise PermissionError("Denied")
    # No audit trail
    # No policy hot-reload
    # Breaks when you add a new action type
    result = await executor.run(action)

# 3 lines. Works for every action type.
from aegis import Action, Policy, Runtime

runtime = Runtime(executor=my_executor, policy=Policy.from_yaml("policy.yaml"))
result = await runtime.run_one(Action("delete", "crm", params={"id": "all"}))
# Policy checked. Approval gated. Audit logged. Done.

Policy (YAML)

0 rules ✅ Valid

Simulate Actions

Custom Action

Type

Target

Params (JSON)

Evaluation Result

Click an action above to see the policy evaluation result

Audit Log

0 entries Import

Audit entries will appear here as you evaluate actions. Try clicking a preset above, then press an action button!

How It Works

⚠️

AI agents without governance are a liability. Uncontrolled agents can delete production data, leak PII, or trigger compliance violations. Aegis adds a policy layer in 2 lines of code — no behavior changes, just safety.

Without Aegis

agent.run(action) # 💥 anything goes

With Aegis

rt.run_one(action) # 🛡️ policy-checked

What happens on each action:

1Agent calls rt.run_one(action)

→

2Aegis matches rules in <1ms

→

3Returns auto / approve / block

→

4Audit entry logged automatically

📝

Write YAML Policy

Define rules for each action type: auto-approve safe reads, require human review for writes, block dangerous operations.

14 presets YAML syntax ~2 min

→

🤖

Agent Sends Actions

Your AI agent (LangChain, CrewAI, OpenAI, etc.) sends each action through Aegis before executing it.

7 adapters 2 lines code ~30 sec

→

🛡️

Instant Decision

Aegis evaluates risk in <1ms: auto-approve, require human approval, or block. Every decision is audit-logged.

<1ms eval 100% audit instant

🤖 AI Agent LangChain / CrewAI / OpenAI

action request

🛡️ Aegis Engine YAML policy + risk eval

decision

✅ auto 🟡 review 🔴 block

Every action is evaluated, logged, and auditable — zero blind spots.

Risk LevelExample ActionDefault Decision

LOWRead contacts✅ Auto-approve

MEDIUMUpdate record🟡 Human review

HIGHBulk export🟡 Human review

CRITICALDelete all data🔴 Blocked

PyPI v0.1.4 518+ tests MIT License Zero runtime deps <1ms eval Type-safe

Works with: LangChain CrewAI OpenAI Anthropic MCP AutoGen any Python agent

$ pip install agent-aegis

version: "1"
rules:
  - name: read_auto
    approval: auto
  - name: write_review
    approval: approve
  - name: delete_block
    approval: block

from aegis import Policy, Runtime

policy = Policy.from_yaml("policy.yaml")
async with Runtime(executor, policy) as rt:
    result = await rt.run_one(action)
    # result.is_allowed → True/False

$ docker run -p 8000:8000 \
  -v ./policy.yaml:/app/policy.yaml \
  ghcr.io/acacian/aegis:latest
# REST API at http://localhost:8000

30s to install

2 min to first policy

0 config files needed

1 dep only PyYAML

No agent changes — wrap existing code, don't rewrite it
Hot-reload policies — update rules without restarting your app
Audit everything — every decision logged for compliance
Type-safe — full type annotations, works with mypy and pyright

See it in action — no install required

Real-World Scenarios

Click any scenario to load its policy and test actions

Why Aegis?

The difference between hoping your AI agent behaves and knowing it does

	Without Governance	With Aegis
Policy changes	Redeploy code	Edit YAML, hot-reload
Risk evaluation	Manual if/else chains	<1ms, declarative rules
Audit trail	Build your own logging	Built-in, compliance-ready
Human approval	Custom workflow code	One-line approval handler
Framework support	Build per framework	7 adapters, one policy
Setup time	Days to weeks	5 minutes

0 actions evaluated in this session

Ready to govern your AI agents?

Add governance to any Python AI agent in 5 minutes. One pip install, one YAML file.

Star on GitHub Read the Docs

Loading Aegis Playground

Aegis Playground

Write a Policy

Simulate Actions

See the Verdict

Policy (YAML)

Simulate Actions

Custom Action

Evaluation Result

Audit Log

How It Works

Write YAML Policy

Agent Sends Actions

Instant Decision

Real-World Scenarios

CRM Agent

Code Agent

Financial Agent

Browser Agent

Data Pipeline

DevOps Agent

Healthcare Agent

E-Commerce Agent

Support Agent

Research Agent

Why Aegis?

Ready to govern your AI agents?