Obscura-then-Chromium fetch + screenshots (tronbrowser.dev)
Maintain this server? Claim it
Call this server's tools from your browser. Nothing to install.
Requests go through mcptry because browsers can't call most MCP servers directly. Header values are sent to the server for that request only and are never stored or logged.
Tool list from our last check. Connect live to refresh it.
Fetch a public web page as markdown, text, links or html. Obscura renders first (fast, light); the ungoogled-chromium engine takes over for JS-heavy pages and bot walls. engine=obscura|chromium pins one. Keyless; a few fetches at a time. 30 s render budget; a page that times out, returns 5xx, or whose host is unreachable is skipped for 30 days with an error starting "FLAGGED until <ISO date>".
Public http(s) URL
Default auto
Default markdown
Cut the answer here. Default 200000, max 400000.
Total render budget in seconds (Obscura, then Chromium). Default and max 30.
This tool isn't marked read-only, so it may change something on the server or in an account you connected.
Pick your client. Values in capitals, like YOUR_TOKEN, are yours to fill in.
Run in a terminal
claude mcp add --transport http tronbrowser https://tronbrowser.dev/mcp/tron
Fetch a public web page as markdown, text, links or html. Obscura renders first (fast, light); the ungoogled-chromium engine takes over for JS-heavy pages and bot walls. engine=obscura|chromium pins one. Keyless; a few fetches at a time. 30 s render budget; a page that times out, returns 5xx, or whose host is unreachable is skipped for 30 days with an error starting "FLAGGED until <ISO date>".
| Argument | Type | Description |
|---|---|---|
| url* | string | Public http(s) URL |
| engine | string | Default auto |
| format | string | Default markdown |
| max_chars | integer | Cut the answer here. Default 200000, max 400000. |
| timeout_s | number | Total render budget in seconds (Obscura, then Chromium). Default and max 30. |
{
"type": "object",
"required": [
"url"
],
"properties": {
"url": {
"type": "string",
"description": "Public http(s) URL"
},
"engine": {
"enum": [
"auto",
"obscura",
"chromium"
],
"type": "string",
"description": "Default auto"
},
"format": {
"enum": [
"markdown",
"text",
"links",
"html"
],
"type": "string",
"description": "Default markdown"
},
"max_chars": {
"type": "integer",
"minimum": 1000,
"description": "Cut the answer here. Default 200000, max 400000."
},
"timeout_s": {
"type": "number",
"description": "Total render budget in seconds (Obscura, then Chromium). Default and max 30."
}
},
"additionalProperties": false
}Render a public web page in the ungoogled-chromium engine and return a PNG of the viewport (1280x900).
| Argument | Type | Description |
|---|---|---|
| url* | string | Public http(s) URL |
{
"type": "object",
"required": [
"url"
],
"properties": {
"url": {
"type": "string",
"description": "Public http(s) URL"
}
},
"additionalProperties": false
}We read every tool's name, description and inputs the way the model will, looking for instructions hidden from you. A clean result means no warning signs, not a guarantee. Scanned Nov 11, 2026.
No readOnlyHint / destructiveHint annotations, so clients can't tell a read from a write without asking.
Fix: Add annotations to each tool, e.g. { readOnlyHint: true } for tools that only read.