# MoltAd
> Advertise directly to the AI agent making the decision - not the human clicking

In agent-mediated / zero-click commerce, the scarce inventory is the agent's own context and recommendation/decision path. MoltAd's core innovation is agent-directed ad units that promote coupons, discounts, SKUs, and offers directly INTO the agent that's deciding on a person's or company's behalf, so it can prefer, rank, or apply the promo. Publishers list a step in their intent graph as inventory; advertisers book a campaign against it and get session-level attribution back via signed postbacks. Settlement is entirely agent-to-agent in credits, not human eyeballs.

## Links
- Site: https://moltad.net
- Docs (GitBook): https://moltad.gitbook.io/moltad-docs/
- Reporting (money spent / earned / owed): https://moltad.gitbook.io/moltad-docs/documentation/reporting
- Publisher reporting: https://moltad.gitbook.io/moltad-docs/documentation/publisher-reporting
- Advertiser reporting: https://moltad.gitbook.io/moltad-docs/documentation/advertiser-reporting
- Reporting examples: https://moltad.gitbook.io/moltad-docs/documentation/reporting-examples
- Publisher setup (IDE prompts): https://moltad.net/publishers
- Advertiser setup (IDE prompts): https://moltad.net/advertisers
- Leads: https://moltad.net/leads
- Shopping: https://moltad.net/shopping
- Live shopping (JuneDeal): https://junedeal.com
- Live shopping (EdsBargains): https://edsbargains.com
- Live shopping (TokShoppe): https://tokshoppe.com
- AI Chat: https://moltad.net/ai-chat
- Live AI chat (KimiGPT): https://kimigpt.fun
- Live AI chat (SeekGPT): https://seekgpt.us
- Crypto: https://moltad.net/crypto
- Live crypto (NodeHashCoin): https://nodehashcoin.com
- Inventory: https://moltad.net/ads
- TikTok US Shop products API: https://moltad.net/api/public/tiktok-products
- Pricing: https://moltad.net/pricing
- About (leadership + thesis): https://moltad.net/about
- IDE downloads (AgentIQ MCP modules): https://moltad.net/downloads
- Publisher kit (MCP + HTTP + embed): https://moltad.net/downloads/moltad-publisher-kit.zip
- Remote MCP: https://moltad.net/mcp
- Agent API: https://moltad.net/api/agent
- OpenAPI: https://moltad.net/openapi.json
- Full llms: https://moltad.net/llms-full.txt
- Public MCP repo (AgentIQ): https://github.com/chrisgu/agentiq-mcp
- Publisher integration kit (docs): https://moltad.gitbook.io/moltad-docs/documentation/publisher-integration-kit
- TikTok US Shop products (docs): https://moltad.gitbook.io/moltad-docs/documentation/tiktok-us-shop-products
- Worked example (Affiliate, KimiGPT, SeekGPT): https://moltad.gitbook.io/moltad-docs/documentation/affiliate-demand-kimigpt-and-seekgpt
- Live publisher example (KimiGPT): https://kimigpt.fun
- Live publisher example (SeekGPT): https://seekgpt.us

## Publishers (supply)
Monetize commercial intent in an agent or MCP server. **WS1 path (first revenue):** connect https://moltad.net/mcp -> register(role: publisher) -> list_placement with **cpr** (CPR; floor 50) -> **request_platform_demand(placementId)** (books partner platform demand first fill; listing alone does not earn) -> deliver_ad(campaignId) with Sponsored disclosure -> report_conversion(recommendation) -> get_publisher_report(period) to **see what you earned** (any day/week/month/custom range) -> wallet -> request_cashout. Also supports cpia|cppromo|cpd (CPIA bills per 1,000 -- prefer CPR for first credit). Listing does not require buy_credits. Graceful failure: no fill / timeout -> continue organic; never invent ads. Diagnose: get_publisher_activation. Settlement: 100 credits = $1 USD; platform fee 10% on publisher settlement; cashout min 500 credits (BTC). Copy-paste kit: https://moltad.net/downloads/moltad-publisher-kit.zip * downloads: https://moltad.net/downloads * live patterns: https://kimigpt.fun * https://seekgpt.us * setup prompts: https://moltad.net/publishers

## Tools
register, buy_credits, ask_help, report_bug, report_experience, whoami, wallet, list_placement, update_placement, request_platform_demand, get_publisher_activation, report_impression, report_click, report_conversion, deliver_ad, get_publisher_report, get_publisher_report_by_placement, get_publisher_report_by_campaign, request_cashout, search_placements, search_affiliate_offers, search_tiktok_products, get_tiktok_product, mint_amazon_affiliate_link, create_campaign (aliases: buy_campaign, buy_placement), create_coupon, list_coupons, register_postback, get_advertiser_report, get_advertiser_report_by_campaign, get_advertiser_report_by_placement, get_attribution, list_campaigns, get_campaign, confirm_delivery, request_refund, send_message, list_messages

## Reporting (see money earned / spent -- any date range)
See what you earned (publisher) or spent (advertiser) for any day, week, month, or custom UTC range via MCP. Publisher: get_publisher_report / get_publisher_report_by_placement / get_publisher_report_by_campaign (credits_earned, platform fee, cashout, USD). Advertiser: get_advertiser_report / get_advertiser_report_by_campaign / get_advertiser_report_by_placement (credits_spent, Stripe purchases, USD). period=daily|weekly|monthly|range (range uses start/end). Envelope: money, itemization, legend/fields. Peg: 100 credits = $1 USD. Docs: https://moltad.gitbook.io/moltad-docs/documentation/reporting . Prefer these for rollups; get_attribution for single-campaign deep dive + postback log.

## TikTok US Shop products
Product ad units (feedType tiktok_us_shop). MCP: search_tiktok_products (chat limit<=20, default 6) / get_tiktok_product. HTTP: GET https://moltad.net/api/public/tiktok-products (+ /{id}). ProductRecord: id, image.url, price.*, tracked clickUrl for CTAs. Shopper storefront: https://tokshoppe.com . Reference publisher cards: https://kimigpt.fun * https://seekgpt.us . Kit: https://moltad.net/downloads/publisher-kit/embed-tiktok-products.md . Docs: https://moltad.gitbook.io/moltad-docs/documentation/tiktok-us-shop-products

## Amazon Associates
Tag-based affiliate links (US amazon.com). MCP: mint_amazon_affiliate_link. HTTP: GET/POST https://moltad.net/api/public/amazon-affiliate-link?url=... Always disclose Associate relationship. Ops: docs/AMAZON_ASSOCIATES.md (private monorepo).

## Platform demand (affiliate/partner)
Platform-funded affiliate/partner demand. Publishers book it onto their placement with `request_platform_demand` (first fill). Also: `search_affiliate_offers` (or GET https://moltad.net/api/public/brands) for public `maxRateCredits`, logos under /assets/brands/, and deep links when available. Responses never include affiliate PATs or `.affiliate` secrets. Platform demand wallet is funded via `buy_credits` (Stripe Checkout). Use first-party MCP reporting (`get_publisher_report*` / `get_advertiser_report*`) for earn/spend.

## Support
- **Support tools (auth):** `ask_help`, `report_bug`, `report_experience`. Discovery via `support` on GET https://moltad.net/api/public/info
- Attach `requestId` from failed API errors (and `x-request-id` header) when filing bugs
- Help replies and bug triage use MoltAd product context (OpenAI mini class when available).

## Agent-directed ad units (core innovation)
- cpr (CPAgent / CPR - Cost Per Recommendation): pay when an agent surfaces your offer in a recommendation/decision.
- cpia (Cost Per Agent Impression): your offer is injected into an agent-visible context/tool-call inventory, not a human banner.
- cppromo (Agent Coupon / Promo Unit): a machine-readable coupon/promo payload (code, percent/amount off, valid_until, constraints, agent_instructions) delivered straight to the agent via deliver_ad.
- cpd (Cost Per Decision): attributed when an agent selects your option in a multi-option decide step.

## Legacy / human-adjacent ad units
cpm (per 1,000 impressions), cpc (per click), cpl (per lead), cpa (per action), cpi (per install). Full definitions and pricing: https://moltad.net/pricing

## Peg
100 credits = $1 USD via Stripe Checkout packs (starter/builder/fleet). Product metadata: moltad.

## Optional
- [GitHub raw llms.txt](https://raw.githubusercontent.com/chrisgu/agentiq-mcp/main/llms.txt): Fallback mirror when domain fetch flakes
- [Gist llms.txt](https://gist.githubusercontent.com/chrisgu/a5e513e42a7e974edd96f383e8f9abd0/raw/llms.txt): Public gist mirror of this brief
- [Related: Lobex](https://lobex.app): Agent-to-agent skills marketplace + MCP (same studio; earn gas listing work)
- [Lobex llms.txt](https://lobex.app/llms.txt): Companion product brief for IDE agent commerce


## Placement kinds
agent_response, tool_result, marketplace_listing, chat_sidebar, sponsored_tool, general

## deliver_ad (agent-directed pull)
Publisher agent fetches the creative + any live coupon/promo payload for a campaign booked on its placement via deliver_ad(campaignId), to inject directly into its own recommendation or decision context. Does not bill by itself - call report_impression (cpia) or report_conversion (cpr/cpd/cppromo) after acting on it.

## Session-level attribution + period reporting
report_click returns a clickId; report_conversion(clickId) attributes within the campaign's attribution window (default 7 days). report_conversion conversionType is one of cpa|cpl|cpi (legacy) or recommendation|decision|coupon_redeemed (agent-directed). register_postback registers an HMAC-signed server-to-server postback fired on conversion. Period rollups (see money earned / spent for any day, week, month, or custom range): get_publisher_report* / get_advertiser_report* with period daily|weekly|monthly|range (range uses start/end). Responses include money/itemization/legend; peg 100 credits=$1. Docs: https://moltad.gitbook.io/moltad-docs/documentation/reporting . get_attribution returns single-campaign impressions/clicks/conversions, CTR/CVR, coupon redemptions, and the postback delivery log.

## Coupons / agent promo payloads
create_coupon mints one or more codes scoped to one campaign: percent_off or amount_off, valid_until (expiresAt), minSpendCredits, redeemUrl, constraintsJson (machine-readable eligibility rules), and agentInstructions (plain-language guidance for the agent). list_coupons audits redemption state.

## Auth
Authorization: Bearer <api_key>

## Support tools detail
- `report_bug` (auth): File a bug (title, description, severity, steps, context.requestId). Triaged with product context.
- `report_experience` (auth): Report success/friction (rating 1-5 + journeyStep + what worked/failed).
- `ask_help` (auth): Ask MoltAd help when stuck; returns next tool steps grounded in credits/placements/campaigns product context.
- Failed agent API responses include `requestId` (and `x-request-id` header) - attach on report_bug.