Pikkola API & Webhook Lab, tool No. 22

The response your browser refuses to show you.

Most APIs answer with headers a page is not allowed to read, and refuse a call that did not come from their own origin. This lab sends it from the edge instead: any method, any header, a JSON body, a bearer token. What comes back is printed whole, timed, folded into a JSON tree, and written out as cURL, fetch, Python or Worker code you can paste.

Send a request one request per press, no account, nothing stored
6
methods
4
code formats
0
requests kept

The lab

Build the call on the left. Read the whole answer on the right.

The request leaves from our edge, not from your tab, so the API sees an ordinary server and answers it. Timeouts are 15 seconds, request bodies are capped at 128 kB and responses are read up to 512 kB. Nothing is retried for you.

POST /api/send, redirects up to 5 hops, private addresses refused

Request GET
URL

Host, Content-Length, Connection and the forwarding headers are set by the edge and cannot be overridden. A User-Agent of your own is allowed.

Response waiting for a request

Type an address and press Send. The status, the time, every header and the whole body land here.

Webhook payload lab

The payload is the part that gets refused at 2 a.m.

Every service has its own shape for the same event, and the shape is what breaks. Pick a preset: it loads the real payload into the body above, sets the content type, and checks the structure as you edit, down to the length limits each platform enforces.

Payload presets

Structure check no preset loaded
  • Load a preset to see which fields the platform requires, which are optional, and which limits apply to this payload.

How it works

A browser is not allowed to ask. A server is.

CORS belongs to the API, not to you.

The other server decides which origins may read its answer, and the browser enforces it before your code sees a byte. A request sent from our worker has no origin to police, so the answer comes back whole, headers included.

Every hop is checked, not just the first.

Redirects are followed by hand, one at a time, up to five. Each target is resolved and refused if it lands on a private, loopback or metadata address, so a public URL cannot be turned into a way into somebody's network.

The tree is for the part you did not expect.

A JSON response is folded by type, so a nested error inside a 200 reads as a branch instead of a wall of text. Any path can be copied as $.data.items[0].id and pasted into jq or your own code.

The code is the request you just made.

cURL, fetch, Python requests and a Cloudflare Worker, generated from the method, URL, headers, auth and body on screen. Copy it into your project and the call behaves the same, minus this page.

Questions

Asked while a token was misbehaving

Why can a browser not call my API from another site?

Because the API's server says so. A cross origin request is sent, and then the browser refuses to hand the answer to your script unless the server sent Access-Control-Allow-Origin for that exact origin and answered the preflight OPTIONS call. Even when it allows the response, most headers stay hidden. Sending the same request from a server avoids all of it, because the rule was written for scripts inside a page.

Where does my request go?

To a Cloudflare Worker, which makes one request to the address you gave and returns the answer to this page. Your token, your body and the address are not written down: no database, no analytics, no request log of ours. The target server sees the request from Cloudflare's network rather than your connection.

Why does an API answer 403 here when it works in my terminal?

Because the call leaves from a shared Cloudflare address with a user agent that names this tool. Sites that block bots, and APIs that limit anonymous calls by IP, may refuse that. Add a User-Agent header of your own in the Headers tab, or send the token the API is asking for. A 403 is an answer too, and the body usually says which rule you hit.

Can it call localhost or an internal service?

No. Each address is resolved first and refused if it sits in a private, loopback, link local or cloud metadata range, and the same check runs on every redirect target. That also means a service that only exists on your laptop cannot be tested from here.

Is there a limit on how much I can send?

A request body of 128 kB, a response read of 512 kB, a 15 second deadline and 120 requests a minute per address. Redirects stop at five. These are the numbers that keep the free tier free and keep the tool from being used as a relay.

Do you keep my tokens or responses?

No. Nothing is written to storage, there is no history, and a response lives in the page until you send the next one. Your language and theme stay in this browser's local storage so the page opens the way you left it, and clearing site data removes them.

Can I send a real Slack or Discord message?

Yes, with a real incoming webhook URL. That is a live call and the message will arrive, so point it at a test channel first. Treat the URL as a secret: anyone who has it can post to that channel.

The request is made. It is not kept.

No account, no analytics and no request log of ours. The target server learns your address only as a Cloudflare address, and the body you sent is the body it received.

Read the privacy note