Vaaya — one prepaid wallet for every paid API your agent needs

Vaaya is a procurement gateway for agents: instead of holding a crypto wallet and a private key (x402 style) or juggling one API key per vendor, your agent talks to a single MCP server — https://vaaya.ai/mcp — and buys capabilities pay-per-call from one prepaid balance.

What that buys you, through one endpoint:

• Image / video / audio generation & editing
• Web search (live, post-cutoff data) and web scraping / crawling
• Document & PDF parsing
• Sandboxed compute (run code in isolated sessions)
• Browser automation
• Email sending and phone calls
• Lead discovery & contact enrichment, market/competitive research
• plus storage, databases, hosting, embeddings, memory, and more

How it differs from the x402 skills you may have seen:

• No crypto wallet, no private key in your env. Auth is OAuth 2.1; billing is a prepaid ledger. Nothing secret lives in your environment for install.
• No per-API signups. One account covers every service in the catalog.
• Billed only on success, per call, in cents — failed calls are never charged — and every call carries a spend ceiling you choose.

Zero-human agent signup (get a working credential + $1 right now)

An agent can open its own account with two POSTs — no human at the keyboard:

1. Register an OAuth client (dynamic client registration, no pre-approval):

   POST https://vaaya.ai/oauth/register
   Content-Type: application/json
   
   { "client_name": "openclaw-agent", "grant_types": ["refresh_token"] }
   

   Do not omit grant_types — the default is the authorization_code grant, which then requires redirect_uris.

2. Create the account with the client_id you just received:

   POST https://vaaya.ai/api/v1/signup/agent
   Content-Type: application/json
   
   { "client_id": "..." }
   

   The response contains an access_token + refresh_token, a $1 starter credit (spendable immediately, expires in 24h), and a claim_url.

3. Store the refresh_token and user_id in your persistent memory or config NOW and reuse them across sessions. Treat the refresh token like a password: keep it in your agent's private state (a file only your process can read, or your client's secret store), and never put it in a shared repo, a chat message, or logs. If you suspect it leaked, revoke the grant at https://vaaya.ai/connected-apps and sign in again. A leaked token only reaches this one Vaaya account, and every call it makes is capped by max_cost_cents and the account's balance. Never re-register: re-registering forfeits your history and grants no new credit.

4. Hand claim_url to your human when they're around. One Google sign-in merges the account into theirs. Once they add a card, a $2 welcome credit line activates — the card settles only what agents actually spend — and connecting GitHub grows the line from their track record. Prepaid credit packs are available at https://vaaya.ai/credits. Claim links expire in 24h — never memorize one; mint a fresh link anytime via GET https://vaaya.ai/api/v1/signup/claim-url (bearer auth). Your tokens keep working across the merge.

Connect via MCP (OpenClaw)

openclaw mcp add vaaya --url https://vaaya.ai/mcp --transport streamable-http --auth oauth
openclaw mcp login vaaya

login opens a browser for the OAuth approval (sign-in with Google). If you signed up agent-first (above), the account already exists — the human claim via claim_url links it to their Google identity. No API keys are ever pasted anywhere; the grant is revocable at https://vaaya.ai/connected-apps.

Usage pattern: consult → use → result

Always start with consult. You never need to know service names, params, URLs, or keys — consult knows the live catalog and hands back the exact call.

1. consult({ intent }) — describe the whole goal in plain English ("make a 10s product demo clip from these screenshots"). It returns either a question to relay to your user (mode: "converse"), or mode: "call" with an ordered list of ready-to-run calls { service, action, params, max_cost_cents, why }.
2. use({ service, action, params, max_cost_cents }) — execute one call exactly as consult handed it to you. Bills on success only. Long-running work returns { async: true, job_id }.
3. result({ job_id }) — poll async jobs. Never re-run use to check on a job — that starts a new, separately-billed job.
4. After the calls finish, call consult again with a one-line outcome for result-aware next steps.

Sandboxes (session / close) bill per second of uptime — always close when done.

Pricing transparency

• Every call is priced in cents, per call, and the price is visible before you run it: consult returns a max_cost_cents with every planned call, and you pass that cap into use — the call can never charge more than the cap.
• Failed calls are never charged. use responses report exactly what was charged (charged_cents) and what remains (balance_remaining_cents).
• The $1 starter credit (agent signup, expires 24h) and any prepaid credit packs ($10 / $30 / $100, at https://vaaya.ai/credits?utm_source=clawhub&utm_medium=agent&utm_campaign=skill) spend first; a card-backed credit line ($2 welcome line plus a GitHub-score line) extends past that. On a payment error (credits_required), do not retry — relay the returned credits_url to your user.
• Full catalog and per-call prices: https://vaaya.ai/services?utm_source=clawhub&utm_medium=agent&utm_campaign=skill
• Agent-readable index: https://vaaya.ai/llms.txt · full tool reference: https://vaaya.ai/llms-full.txt