A multi-format Data Doctor and viewer for 22 formats — from JSON and CSV to Parquet, Excel, AI model files, and X.509 certificates — that also publishes payloads as mock REST API webhooks.
The three capabilities that define Yellorn.
29 auto-fix passes repair broken JSON, XML, YAML, and CSV from ChatGPT, Excel exports, or truncated logs — markdown fences, trailing commas, smart quotes, BOMs, Python literals. Every fix is shown in the status bar; nothing is re-shaped silently.
Click Publish and any payload becomes a global HTTPS endpoint in seconds. The URL accepts every HTTP method; only you can see the request log. Set the response status code (200 / 401 / 429 / 500) and custom headers to test failure paths — and on paid tiers, pick a memorable custom path.
Convert between JSON, XML, YAML, CSV, and Python in one click. Lossy conversions warn before re-shaping, and a single Cmd+Z reverts both the text and the format.
Smaller features that turn the trio into a daily-driver workbench.
Open 22 formats in one editor. Drop a binary file — Parquet, Arrow, Excel, MessagePack, CBOR, Protobuf, BSON, gzip — an AI model (Safetensors, GGUF, ComfyUI PNG), or a certificate (PEM, PFX, CRT, CSR) and Yellorn decodes it locally to a browsable preview. No upload.
Make your mock responses come alive. Wrap a Python-flavoured expression in {{ }} and it renders fresh on every request — {{ uuid4() }}, {{ now() }}, {{ randint(1, 100) }}, {{ choice(...) }}, or echo the caller with {{ request.ip }}. A token picker, autocomplete, and live preview are built in.
Explore data as a tree, a node graph, or a sortable table with inline cell editing. Each format picks the views that fit it.
Query your data live with JMESPath — like `users[?age > `30`].name` — with match counts and prev/next navigation.
Send GET / POST / PUT / PATCH / DELETE requests from the browser and save templates for reuse. SSRF protection and a loop guard prevent abuse.
Four levels of cleanup: Smart (full recovery) → Simple (preserve types) → Safe (layout only) → Compress. The status bar shows which one ran.
Manually back up your open tabs to the cloud and restore them on any other signed-in browser. The Back up / Restore controls and a quota-aware usage bar live in Settings; tabs over your size quota are skipped during back up.
Select two tabs and run them through Monaco's read-only diff editor. JSON is key-sorted before diffing so key reorderings don't pollute the output — only real structural differences show up.
22 formats in one editor — paste text or drop a file. Binary and certificate files decode locally; nothing is uploaded to view it.
Edit, fix, convert, and visualize. Each ships its own best-effort recovery ladder.
JSON · JSON Lines (NDJSON) · XML · YAML · TOML · CSV · ASCII Table · SQL · Python · Raw Text
Open and inspect — decoded locally in a worker to a browsable JSON preview, no upload.
Apache Parquet · Apache Arrow / Feather · Excel (xlsx / xlsm / xls) · MessagePack · CBOR · Protobuf · BSON · gzip
Read headers and embedded metadata without touching the multi-GB payload.
Safetensors (tensor header) · GGUF (llama.cpp model metadata) · ComfyUI (workflow from a generated PNG)
The X.509 PKI family, decoded to a JSON insight — expiry, algorithm, fingerprint, warnings. Private-key material is never shown.
PEM · DER · PKCS#12 / PFX (.pfx / .p12) · PKCS#7 (.p7b / .p7c) · PKCS#10 CSR · public / private keys
What people actually paste into Yellorn — and how it helps.
Paste JSON returned by ChatGPT or Claude. Markdown fences and explanatory comments are stripped automatically; Python-style None / True / False are normalised; smart quotes from rich-text pastes are replaced.
Point Stripe / GitHub / Polar / Slack at a yellorn URL while you build. Watch the request log fill in real time; copy any payload back into the editor for fast iteration.
Per-environment URLs with configurable response status codes — exercise your client's failure paths without redeploying anything.
Paste broken JSON from a Zendesk ticket or a server log. The recovery hints show what the customer actually meant before you write a fix.
Drop a .parquet, .arrow, .xlsx, .msgpack, .cbor, .pb, or .bson file and read it as a tree or table — decoded locally, because the file never leaves your browser.
Read a .safetensors or .gguf header — tensor shapes, dtypes, parameter count — without downloading gigabytes of weights. Or drop a ComfyUI PNG to recover its full prompt graph.
Publish a payload whose fields come alive on every request: {{ uuid4() }} for ids, {{ now() }} for timestamps, {{ randint() }} / {{ choice(...) }} for varying values, and {{ request.* }} to echo the caller. Point your app at the URL and develop against responses that look real.
Drop a .pem, .crt, .pfx, or .csr and Yellorn answers the questions you actually have: is it expired (or expiring within 30 days)? Who is it for? Is it self-signed or using a weak key / SHA-1 signature? What's the SHA-256 fingerprint? Encrypted PFX bundles prompt for a password and unlock in the browser — decoded locally, private-key material is never shown.
UTF-8-BOM CSVs from Excel, smart-quoted YAML from Word templates, tab-indented YAML from auto-formatters — all repaired transparently.
Open the editor, paste a payload, and try Smart Format. Sign-in is optional until you publish a webhook.