AgentEazy

Turn any Python utility into a callable AI tool in one command.

Your code becomes a live API that any AI agent can find, call, and pay — without rewriting a single line.

Website · Dashboard · Registry · Docs

---

Try it right now

No install needed. Call a live agent from your terminal:

curl -s -X POST https://simondusable--agenteazy-gateway-serve.modal.run/agent/zxcvbn-python/ \
  -H "Content-Type: application/json" \
  -d '{"verb":"DO","payload":{"data":{"password":"monkey123"}}}' | python3 -m json.tool

{
  "status": "completed",
  "output": {
    "score": 1,
    "feedback": {
      "warning": "This is a very common password.",
      "suggestions": ["Add another word or two. Uncommon words are better."]
    }
  }
}

That's a real GitHub repo (zxcvbn-python), auto-wrapped and live. 18+ more in the registry.

---

Deploy your own agent

pip install agenteazy
agenteazy signup your-github-username --email you@example.com
agenteazy deploy github.com/you/your-repo

Done. Your repo is live. Every AI agent on earth can now find and call it.

What happens under the hood

1. Analyze — Clones the repo, parses the AST, detects the main public API function
2. Wrap — Generates a FastAPI wrapper that handles imports, dependencies, and dispatch
3. Upload — Pushes to serverless infrastructure (idle agents cost $0)
4. Register — Adds to the public registry so other agents can discover it

---

Python SDK

from agenteazy import AgentEazy

client = AgentEazy()

# Discover agents
agents = client.find("password strength")

# Call agents
result = client.do("zxcvbn-python", {"password": "monkey123"})
# → {"score": 1, "feedback": {"warning": "This is a very common password."}}

result = client.do("dateparser", {"date_string": "tomorrow at 3pm"})
# → "2026-03-17T15:00:00"

result = client.do("langdetect", {"text": "Bonjour le monde"})
# → "fr"

result = client.do("python-ftfy", {"text": "émile dupont"})
# → "émile dupont"

result = client.do("autopep8", {"source": "x=   1\nif  x==1:\n  print( 'hello' )"})
# → "x = 1\nif x == 1:\n    print('hello')\n"

---

LangChain Integration

Every agent becomes a LangChain BaseTool:

pip install agenteazy[langchain]

from agenteazy.integrations.langchain import AgentEazyTool, AgentEazyToolkit

tool = AgentEazyTool.from_agent("zxcvbn-python")
result = tool.invoke({"password": "test123"})

# Discover multiple tools from the registry
toolkit = AgentEazyToolkit()
tools = toolkit.get_tools(query="text processing", limit=5)

---

CrewAI Integration

pip install agenteazy[crewai]

from agenteazy.integrations.crewai import AgentEazyCrewTool

tool = AgentEazyCrewTool.from_agent("zxcvbn-python")

---

Live Agent Registry

18 agents you can call right now, all auto-wrapped from real GitHub repos:

Agent	What it does	Try it
zxcvbn-python	Password strength scoring	{"password":"monkey123"} → score 1/4
langdetect	Detect text language	{"text":"Bonjour"} → "fr"
dateparser	Parse natural language dates	{"date_string":"tomorrow at 3pm"} → datetime
python-ftfy	Fix broken Unicode	{"text":"émile"} → "émile"
autopep8	Format Python to PEP 8	{"source":"x= 1"} → "x = 1\n"
validators	Validate emails, URLs, IPs	{"value":"test@example.com"} → true
python-slugify	Text to URL slug	{"text":"Hello World!"} → "hello-world"
emoji	Emojize/demojize text	{"string":":thumbs_up:"} → 👍
xmltodict	XML to JSON dict	{"xml_input":"<a>1</a>"} → {"a":"1"}
mistune	Fast Markdown → HTML	{"s":"**bold**"} → "<strong>bold</strong>"
dateutil	Parse date strings	{"timestr":"March 5th 2024"} → datetime
arrow	Date/time with timezone	{} → current datetime
prettytable	Parse HTML tables	HTML → structured rows
python-markdown2	Markdown → HTML	{"text":"# Hello"} → "<h1>Hello</h1>"
python-markdownify	HTML → Markdown	{"html":"<h1>Hello</h1>"} → "# Hello"
num2words	Numbers to words	{"number":42} → "forty-two"
shortuuid	Generate short UUIDs	{} → "N6nquzbtjAF..."
humanize	Natural language lists	{"items":["Alice","Bob"]} → "Alice and Bob"

Browse all at agenteazy.com/agents.

---

What repos work best

AgentEazy auto-detects the right entry point for stateless Python utilities — functions that take input and return output.

Works great	Examples
Text processing	slugify, ftfy, emoji, markdown2
Validation & parsing	validators, dateparser, dateutil
Security tools	zxcvbn (passwords), nh3 (HTML sanitization)
Code formatting	autopep8
Data conversion	xmltodict, markdownify, num2words
Language detection	langdetect

Needs --entry flag	Why
Large repos with many functions	Auto-detect may pick a helper instead of the main API
Class-based APIs	Specify --entry "file.py:Class.method"

Not a good fit	Why
HTTP clients (requests, boto3)	They make outbound calls, not pure functions
Frameworks (Flask, Django)	Already have their own endpoints
Heavy ML models	Cold start too slow for serverless

Override the entry point when needed:

agenteazy deploy github.com/you/repo --entry "mypackage/core.py:process"
agenteazy deploy github.com/you/repo --entry "mypackage/model.py:MyClass.predict"

---

TollBooth — Agent Credits

Agents can charge credits per call. Set a price and the gateway handles billing.

agenteazy deploy github.com/you/your-repo --price 10

• Free by default — no credits needed to call free agents
• You set the price — 1 credit, 10, 100 — your choice
• 80/20 split — you keep 80%, platform takes 20% for infrastructure
• 50 free credits on signup, buy more at agenteazy.com/dashboard

---

AgentLang — 10 Universal Verbs

One protocol for all agents. One HTTP POST, one envelope:

POST /agent/{name}/
{"verb": "DO", "payload": {"data": {"input": "hello"}}}

Verb	What it does	Status
DO	Execute the agent's main function	Live ✅
ASK	Query capabilities and parameters	Live ✅
FIND	Search the registry for agents	Live ✅
PAY	Transfer credits between agents	Live ✅
SHARE	Pass context between calls	Live ✅
REPORT	Get audit log	Live ✅
STOP	Halt a running task	Live ✅
WATCH	Subscribe to events	Planned
TRUST	Authenticated sessions	Planned
LEARN	Ingest new knowledge	Planned

---

CLI Reference

# Deploy
agenteazy deploy <repo>                    # Full pipeline
agenteazy deploy <repo> --local            # Test locally first
agenteazy deploy <repo> --price 10         # Paid agent
agenteazy deploy <repo> --entry "f.py:fn"  # Override entry point
agenteazy deploy <repo> --env KEY=VALUE    # Inject env vars

# Analyze & Wrap
agenteazy analyze <repo>                   # Inspect without deploying
agenteazy wrap <repo>                      # Generate wrapper only
agenteazy batch-deploy <repos.txt>         # Deploy from file

# Account
agenteazy signup <username> --email <email>  # Get API key + 50 credits
agenteazy balance                            # Check credits
agenteazy transactions                       # View history

# Registry
agenteazy search "query"                   # Find agents
agenteazy list                             # List all agents
agenteazy status                           # Health check

---

Architecture

agenteazy deploy <repo>
    │
    ├── ANALYZE  → Clone, parse AST, detect public API function
    ├── WRAP     → Generate FastAPI wrapper with full import handling
    ├── UPLOAD   → Push to serverless volume (idle = $0)
    ├── REGISTER → Add to searchable registry
    └── LIVE     → Callable at gateway/agent/{name}/

Gateway (serves all agents)  ←→  Registry (search + credits)
    │                                    │
    └── Agent Volume                     └── TollBooth
        /agents/zxcvbn-python/               Signup, balance,
        /agents/langdetect/                  deduct, earn, transfer
        /agents/dateparser/
        ... (18 agents)

---

Contributing

Best first contribution: OpenAI Function Calling integration. Same pattern as the LangChain wrapper — one file, self-contained. See agenteazy/integrations/langchain.py as a template.

git clone https://github.com/SimonSliman/agenteazy
cd agenteazy
pip install -e ".[dev]"
pytest tests/ -q  # 20 tests

---

License

MIT