lulu-ads 0.1.0

Disclosed sponsored slots for MCP servers and AI agent tools. Ships data, never directives. Fail-open by design.

lulu-ads

Monetize your MCP server or agent tool with one labeled sponsored line.

Lulu Ads attaches a disclosed, labeled data field to your tool responses. The host model (Claude, Cursor, any agent) decides on its own judgment whether it's relevant enough to surface — we never instruct a model to display anything. If it renders and gets clicked, you earn 70% on CPA. If it doesn't — nobody pays, nothing breaks.

Zero-friction start (MCP)

Add the Lulu Ads MCP server and ask your agent to do the rest:

claude mcp add --transport http lulu-ads https://ads.getlulu.dev/mcp

Then just say:

monetize my server

The agent will fetch the right integration guide for your stack, register a publisher (with your consent), wire up the one-liner below, and verify it produced a live slot.

Quickstart (Python)

pip install lulu-ads

# pip install lulu-ads
from lulu_ads import LuluAds
ads = LuluAds(publisher_id="pub_123", api_key="lk_...")

result = search_flights("TLV", "BKK", dates)
result["sponsored"] = await ads.sponsored_slot(
    context={"tool": "search_flights", "category": "travel.flights"},
    timeout_ms=150,
)
return result

One line for FastMCP servers — credentials come from the environment, the call itself takes no arguments:

export LULU_ADS_PUBLISHER_ID=pub_123
export LULU_ADS_API_KEY=lk_...

mcp.add_middleware(LuluAdsMiddleware())

Quickstart (TypeScript)

npm install @getlulu/ads

// npm install @getlulu/ads
import { LuluAds } from "@getlulu/ads";
const ads = new LuluAds({ publisherId: "pub_123", apiKey: "lk_..." });
result.sponsored = await ads.sponsoredSlot({ context: { tool: "search_flights" } });

Guarantees (enforced in code, not just promised)

Guarantee	How
A tool call can never break because of ads	every failure path returns None/null; hard 150ms timeout
Always disclosed	label: "Sponsored" is set by the SDK and cannot be removed
No prompt injection, ever	we ship a data field; there is no display instruction anywhere
No PII leaves your server	context accepts an allowlisted key set only
Quality-gated	every creative passes Dali scoring (≥70) before it can fill a slot
Intent, not identity	targeting uses the session's stated intent; no user profiles exist
Misconfigured? Still safe	missing creds → client is inert, returns nothing, never crashes

Framework adapters

Full details, exclude-lists, and every supported stack: docs/integrations.md.

LangGraph / LangChain (langchain>=1.0) — middleware, no manual attach step:

from langchain.agents import create_agent
from lulu_ads.integrations.langchain import LuluAdsAgentMiddleware

agent = create_agent(model, tools, middleware=[LuluAdsAgentMiddleware()])

CrewAI (crewai>=1.9.1) — one call at startup, before you kick off a crew:

import lulu_ads.integrations.crewai as lulu_crewai
lulu_crewai.install()

TypeScript MCP servers — wrap the server once, before registering tools:

import { withLuluAds } from "@getlulu/ads";
withLuluAds(server);

Runtime owners (chat bots, WhatsApp/Telegram agents, anything that owns the final message the user sees) — append a disclosed suffix after generation, as harness code, never as a model instruction:

from lulu_ads import format_suffix
final_message = model_output + format_suffix(sponsored)

Get a publisher ID

Three ways:

• Ask an MCP-connected agent to call the create_publisher tool on https://ads.getlulu.dev/mcp
• Early beta signup: https://getlulu.dev/publishers
• POST https://ads.getlulu.dev/publishers with {"name", "contact_email", "server_url"}

See docs/quickstart.md for the full walkthrough and docs/contract.md for the wire-level API.

Docs: https://getlulu.dev/docs · Built by Lulu · Quality gate: Dali