Agent-Aegis Playground - AI Agent Governance in Your Browser

Initializing Python runtime...

Tip: Agent-Aegis evaluates policies in under 1ms

Click "Run aegis scan" to analyze the code...

guardrails:
  pii:
    action: mask        # mask | block | warn | log
    categories:
      - email
      - credit_card
      - ssn
      - korean_rrn
      - api_key
  injection:
    action: block       # block | warn | log
    sensitivity: medium  # low | medium | high

integrations:
  patch_openai: true    # auto-patch OpenAI client
  patch_anthropic: true # auto-patch Anthropic client

audit:
  backend: sqlite       # sqlite | redis | postgres

$ python

>>> import aegis

>>> aegis.auto_instrument()

# Option A: Two lines of Python
import aegis
aegis.auto_instrument()

# Option B: Zero code changes — just an env var
$ AEGIS_INSTRUMENT=1 python my_agent.py

# Fine-grained control
aegis.auto_instrument(
    frameworks=["langchain", "openai_agents"],  # specific frameworks only
    on_block="warn",       # "raise" (default) | "warn" | "log"
)

LangChain CrewAI OpenAI Agents SDK OpenAI API Anthropic API LiteLLM Google GenAI Pydantic AI LlamaIndex Instructor DSPy

Guardrail	Default	Catches
Prompt injection	Block	85+ patterns, multi-language (EN/KO/ZH/JA)
PII detection	Warn	12 categories (email, credit card, SSN, API keys…)
Prompt leak	Warn	System prompt extraction attempts
Toxicity	Warn	Harmful/abusive content (opt-in to block)

Your code                          Agent-Aegis layer (invisible)
─────────                          ───────────────────────
chain.invoke("Hello")       ──▶  [input guardrails] ──▶ LangChain ──▶ [output guardrails] ──▶ response
Runner.run(agent, "query")  ──▶  [input guardrails] ──▶ OpenAI SDK ──▶ [output guardrails] ──▶ response
crew.kickoff()              ──▶  [task guardrails]  ──▶ CrewAI     ──▶ [tool guardrails]   ──▶ response

0 Evaluated

0 Auto-approved

0 Needs Approval

0 Blocked

- Avg Latency

📝

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)

# 2 lines. Works for everything.
import aegis
aegis.auto_instrument()  # auto-patches all frameworks + activates guardrails

# That's it. OpenAI/Anthropic calls are now governed.
# PII masked, injections blocked, every decision audited.

Custom Action

Type

Target

Params (JSON)

Click an action above to see the policy evaluation result

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

Budget $10.00

Warn Threshold 80%

Soft Limit 90%

Model

Input Tokens

Output Tokens

EU AI Act NIST AI RMF SOC2 ISO 42001 OWASP Agentic Top 10

1 Your agent searched and found options

2 Agent showed you these options:

3 What your agent DIDN'T show you:

Selected Option SELECTED

query_database — Read customer records from CRM

impact: 0.1 | target: crm_database

Action Type

Target

Agent's Declared Impact (6D vector)

Low High

Low = only obvious attacks · Medium = known attack patterns · High = aggressive detection (may flag benign text)

How It Works

⚠️

AI agents without governance are a liability. Uncontrolled agents can delete production data, leak PII, or trigger compliance violations. aegis.auto_instrument() adds full security in 1 function call — no behavior changes, just safety.

Without Agent-Aegis

agent.run(action) # 💥 anything goes

With Agent-Aegis

aegis.auto_instrument() # 🛡️ everything governed

What happens on each action:

1Call aegis.auto_instrument() once at startup

→

2Injection + PII scan + rule match — 2.65ms total (0.5% of LLM latency)

→

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 Agent-Aegis before executing it.

7 adapters 2 lines code ~30 sec

→

🛡️

Instant Decision

Agent-Aegis runs 4 guardrail scans + risk eval in 2.65ms (0.5% of a typical LLM call). Auto-approve, review, or block — every decision audit-logged.

2.65ms / call 100% audit instant

🤖 AI Agent LangChain / CrewAI / OpenAI

action request

🛡️ Agent-Aegis Engine YAML policy + risk eval

decision

✅ auto 🟡 review 🔴 block

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

LOWRead contacts✅ Auto-approve

MEDIUMUpdate record🟡 Human review

HIGHBulk export🟡 Human review

CRITICALDelete all data🔴 Blocked

PyPI v0.7.0 2540+ tests MIT License Zero runtime deps 2.65ms / 4 scans 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

import aegis
aegis.auto_instrument()  # auto-patches all frameworks, activates everything

# OpenAI/Anthropic calls are now governed.
# PII masked, injections blocked, all audited.

$ 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 Agent-Aegis?

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

	Without Governance	With Agent-Aegis
Policy changes	Redeploy code	Edit YAML, hot-reload
Risk evaluation	Manual if/else chains	2.65ms with guardrails, 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.