-
Notifications
You must be signed in to change notification settings - Fork 28
Decide
Payment wires: every Decide route accepts x402 and MPP on the same 402, and prepaid card credits. See Paying with x402 and Paying with MPP.
Describe a job in plain language and get back a call-ready plan: which tools, in what order, with fallbacks and parameters that validate against each tool's schema. The plan covers this catalog and outside x402 sellers that answered a live 402 in the last 7 days. Run the plan yourself, or send it back to Agent402 to run it, and the decision fee returns as credit.
The live page, with the current price per depth: https://agent402.tools/decide
| Route | MCP tool | What it does |
|---|---|---|
POST /api/decide |
decide.plan |
Returns a decision: a plan, cost and latency estimates, a confidence score and any gaps. |
POST /api/decide/execute |
decide.execute |
Runs a decision's plan within your budget and returns each step's result with its receipt. |
POST /api/decide/feedback |
decide.feedback |
Free. Reports whether one step worked; feeds future rankings within a bounded range. |
curl -X POST https://agent402.tools/api/decide \
-H "Content-Type: application/json" \
-d '{"task":"Resolve the ENS name vitalik.eth and list its token balances on Base","depth":"plan"}'The first call answers 402 with the price for that depth; pay it with any x402 or MPP client and repeat.
-
quick: the single best tool for the whole job, with fallbacks. -
plan: the job split into steps, each with a primary tool, fallbacks and dependencies. -
full: the plan plus params filled from your task, a compiled prompt to run it, and cost and latency estimates.
Each step names its tool, endpoint, seller, price, input params and fallbacks, and says whether the tool is ours (firstParty). Values the task gives are filled in. A value the task does not give is named as a <placeholder> and listed in exampleParamsNeedInput, so the agent knows what to supply.
Chained steps. When a later step needs something an earlier step produces (an address an ENS lookup resolved, say), the plan writes {{step N}} for it and lists the earlier step in dependsOn. An agent replaces it with the matching field from that step's response. When the plan runs through /api/decide/execute, Agent402 fills it in itself: from a field with the parameter's name, or the one address or IP the earlier step returned. If the value cannot be named without guessing, the step is skipped and nothing is paid for it.
One formula for every seller, ours included: fit to the step, observed reliability, price, schema quality and a freshness pass mark. The weights are the same for every tool and the formula has no term for who sells it. The model that judges fit sees the same bounded description for every tool, and outside listing text is treated as data, never as instructions.
- The decision fee comes back as a credit worth 100% of it, valid 24 hours, toward running that plan.
- Our tools run at list price. Outside tools are paid on your behalf at the seller's price plus a disclosed 5% routing fee.
- Spend stops at your budget, fallbacks run in order, and a run where no step succeeds is not charged.
- Every step returns its own receipt; outside results are marked as untrusted content.
- x402 Index and Router: the seller index Decide plans over.
-
MCP Connector:
decide.plan,decide.executeanddecide.feedbackon the hosted connector. - Tool Catalog
agent402.tools - the applied layer of Agentic Finance: 500+ pay-per-call tools for AI agents over x402 and MPP (Paying with x402 · Paying with MPP · Glossary) · synced from wiki/ in the main repo - edit there, not here.
Using it (for agents / buyers)
- Getting Started
- Agentic Finance
- Why Pay Here
- Reports, Monitors and Credits
- Paying with x402
- Paying with MPP
- Robinhood Chain (USDG)
- Paying with Compute
- MCP Connector
- Adapters
- AWS Bedrock AgentCore
- Decide - describe a job, get a plan
- Tool Catalog
- Skill Packs
- x402 Index and Router
- x402 Leaderboard
- LLM Gateway (OpenAI /v1)
- LLM Proxy Gateway
- Image Generation Gateway
- Code Execution Sandbox
- Text-to-Speech
- Speech-to-Text
- Text Embeddings
- Payments and x402
- Memory and Coordination
Tollbooth (for site owners)
- Pay-per-crawl - what it is, install, modes
- Pay-per-crawl Walkthrough - 5-min hands-on
- Tollbooth for Agencies - many-site playbook
- Try Tollbooth Cloud (managed)
Website & Developer
- Reports · Monitors · Credits - buy by card
- Quickstart - first call in 60 seconds
- Playground - try tools in your browser
- SDK REPL - live code editor
- API Explorer - browse OpenAPI
- Adapter Docs - per-framework guides
- Workflows - chaining patterns
- Blog · Changelog
Under the hood