Discovery

LLMs.txt

PayWeave serves llms.txt and llms-full.txt files — a plain-text markdown convention designed for LLM consumption. These files give language models a concise, human-readable description of all available skills, their pricing, and how to invoke them, without requiring the model to parse structured JSON or OpenAPI specs.

URLs

File	Global URL	Workspace URL
llms.txt	/llms.txt	/:workspaceId/llms.txt
llms-full.txt	/llms-full.txt	/:workspaceId/llms-full.txt

Both files return Content-Type: text/markdown; charset=utf-8 and are cached for five minutes.

Summary vs full

llms.txt is a concise summary — suitable for including in a system prompt or context window. It lists each skill's name, type, base URL, OpenAPI URL, and endpoint paths with pricing.

llms-full.txt extends the summary with per-endpoint detail: description, input schema, example input, and example output. Use the full variant when the model needs to understand exactly how to call an endpoint.

File structure

Plain Text

# PayWeave

> Payment-enabled platform for monetizing APIs, serverless functions,
> and file downloads via the Machine Payment Protocol (MPP).

PayWeave lets developers monetize their APIs with per-call pricing.
Agents and LLMs can discover available skills and invoke them with
micropayments.

## Skills

### Sentiment Analysis API (gateway)
Analyse text sentiment using a fine-tuned model.
- Base URL: `https://api.payweave.dev/gw/gw_abc123`
- OpenAPI: `https://api.payweave.dev/gw/gw_abc123/openapi.json`
- Endpoints:
  - POST `/v1/analyse` — Analyse Text ($0.0050/call)

### Image Classifier (function)
Classify images into 1000 ImageNet categories.
- Base URL: `https://api.payweave.dev/fn/fn_xyz789`
- OpenAPI: `https://api.payweave.dev/fn/fn_xyz789/openapi.json`
- Endpoints:
  - POST `/classify` — Classify Image ($0.0100/call)

## How to Use

1. Discover skills at `/api/discovery/skills`
2. Each endpoint requires payment via MPP (Machine Payment Protocol)
3. Send requests to the base URL + endpoint path
4. Include MPP payment credentials in the request

## Links

- Skills API: `/api/discovery/skills`
- AI Plugin: `/.well-known/ai-plugin.json`
- Full details: `/llms-full.txt`

Additional sections in llms-full.txt

The full variant appends endpoint-level detail for each entry:

Plain Text

  - POST `/v1/analyse` — Analyse Text ($0.0050/call)
    Sentiment score between -1.0 (negative) and 1.0 (positive).
    - Input schema: `{"text":{"type":"string"}}`
    - Example input: `{"text":"PayWeave is great!"}`
    - Example output: `{"score":0.92,"label":"positive"}`

Include /llms.txt in an agent's system prompt or context to give it instant awareness of all available paid skills on the platform. Use /llms-full.txt when the agent needs input schemas or examples to call endpoints correctly.

Referrals section

Both files include a Referrals section explaining that agents can earn referral rewards by appending ?ref=WALLET_ADDRESS to any payment URL. The referral fee is deducted from the workspace owner's share; the payer always pays the same total amount.

Refunds section

Both files include a Refunds section noting that failed requests (5xx errors, function execution errors, file not found) trigger an automatic refund back to the payer address via on-chain ERC-20 transfer.

Skills API MCP Server