Your agent needs a search, a page or a contact. It stops, you create an account, you paste a key. We are forming a group of people who work with an agent all day (Claude Code, Codex, Cursor and others) to ask providers for free quota, reachable through one MCP server in place of one account per tool. After the free quota, you pay per call.
Under a minute. No card. For now, joining is an email: tell us your agent and your tools. Sign-in with GitHub, Google or email is coming.
Free quota first. Then pay per call.
Signing up takes a person: a form, an email check, a key to copy. So the agent stops and waits for you, tool after tool. And one person can't ask a provider for more than the standard tier. A group that shows up with its real usage can ask.
Tell us which tools your agent uses and the ones you want free quota on.
When enough members ask for a tool, we send its provider the group's numbers and ask for free quota. We publish the answer: yes, no, or no reply.
Through one MCP server you add to your agent.
This board will show, for each tool, how many members asked for it, and every provider's answer. One member, one vote per tool. It fills in as members join.
| Category | Tool | Requests | Status |
|---|---|---|---|
| Web search | Search tool A | 72% | Collecting requests |
| Web search | Search tool B | 48% | Collecting requests |
| Page extraction | Extraction tool | 100% | Letter sent |
| Contact and company data | Data tool | 100% | Yes: free quota |
| Voice | Voice tool | 20% | Collecting requests |
A tool on this board is not a partner. Its provider may never have heard of us.
A vote says you want a tool. A count says how much you use it. Providers answer counts. The usage report is coming soon; here is what it will do.
$ npx zando-usage --dry-run Last 30 days, Claude Code web search (tool 1) 412 calls page extraction (tool 2) 187 calls contact data (tool 3) 23 calls Nothing was sent. Run without --dry-run to send these three lines.
Example output. It reads the sessions your agent already keeps on your machine and counts the calls per tool. That is all that will ever leave your machine: tool names and counts. No prompts, no code, no file names, no paths. The code will be public.
Optional. Claude Code first; Codex and Cursor next. It counts tool calls made by the agent, not calls made by your own apps.
Bringing people. Every person who joins and votes for the tools you want moves them closer to the point where we write to the provider.
Or running an agent in production? Free quota won't change much for you. Tell us what you buy and from whom, and we will take your volume to those providers for a group price.
[email protected]If no provider says yes, you get nothing, and you will have lost a minute. If they do, free quota runs out, and after it you pay per call. You never pay to join.
You keep your accounts and keys. The group adds free quota on tools you haven't set up yet, and a way to try a competitor of a tool you pay for without opening another account.
When the MCP server exists, your calls will go through it. We will see which tool was called, when, and whether it worked. We will not keep the content of requests or results unless you agree. We will publish the exact list before you connect.
Your agent can name one. If it doesn't, we pick the cheapest one that returns a good result, and we tell you which. A provider can't pay to be picked.
Your email, the agent you use, the tools you picked, and your counts if you send them. Providers see totals for the group. They never get your name or your email. You can leave: ask, and we delete it.
They already give free quota to win users. We bring them people with counted usage, in one conversation. Some will say no. You will see which ones.
The provider's public price, not more. The plan is that providers pay us, not you. This is not final, and we will publish it before you spend anything.
No date we can't keep. The board shows how far each tool is from the point where we write to its provider, and every answer is posted there.