browdenCAUTION
Read-only browser MCP server — a local, configurable MCP shell around real Chrome for LLM agents.
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
[](https://github.com/nishantsny/browden/actions/workflows/e2e.yml) [](./LICENSE) [](https://www.python.org/downloads/) [](https://modelcontextprotocol.io)
A local, cross-platform, read-only (configurable) MCP shell around a real Chrome browser. It lets an LLM agent look at and navigate the web through your own browser. The agent can read pages, query the DOM, and take screenshots, but can never execute any write action. The MCP is configurable to allow button-clicks and text-fill, allowlisted per website and visible element.
Short demo video: https://youtu.be/q-W3Z9nlj58
Disclaimer
With great power comes great responsibility: Only point this MCP to sites whose Terms of Service permit automated access.
When to use browden
browden is deliberately narrow: safe, local, undetected, and read-only, with allowlisted writes. This allows your agent to run wild on your own logged-in Chrome. Use browden for your daily research needs + a few writes. Defer to richer automation tools when you need to drive the browser rather than read it.
Typical usecases:
- Let an agent read and navigate your logged-in pages while it stays
structurally unable to click "Buy", send mail, or delete anything.
- You need a prompt-injection perimeter around the agent.
When NOT to use browden
a9f16e72059fOBSERVED · 2026-10-11Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control.
claude mcp add browden -- uvx browden
{
"mcpServers": {
"browden": {
"command": "uvx",
"args": [
"browden"
]
}
}
}Exposed tools (2)
2 read · 0 write · 0 destructive.
| Tool | Risk | Description |
|---|---|---|
list_tabs | read | List all open browser tabs across all profiles |
new_blank_tab | read | Open a new blank tab and return it (navigate it afterwards). |
Trust audit
CAUTIONgrade B · trust 87/100 Install with care. The audit found things worth knowing before you trust its output.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | WARN |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- none-observed
- Network
- declared (9 observation(s))
- Shell
- none-observed
- Dependencies
- pinned
- Secrets in source
- none-found
Findings (12)
f"--remote-allow-origins=http://127.0.0.1:{port}",url = f"http://127.0.0.1:{port}/json/version"page_url = f"http://127.0.0.1:{httpd.server_address[1]}/"service_name: {"type": "sse", "url": f"http://127.0.0.1:{port}/sse"}print(f"\n{'phase':<16}{'n':>4}{'uss_mb':>9}{'Δidle':>8}"tranco-mini.txt.gz
(tmp_path / "id_rsa").write_text("PRIVATE KEY")secret = tmp_path / "id_rsa"
secret = tmp_path / "id_rsa"
assert "id_rsa" not in sent
await server.upload_file("#bill_file_expense", str(receipts / "id_rsa"), "h1")"url": "http://127.0.0.1:22001/sse"
Gates applied: no_behavioural_pass.
a9f16e72059ffull audit observations/trust-audit/mcp-server/nishantsny__browden.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-11 | a9f16e72059f | CAUTION | B | 87 | first audit |
Questions
What is the browden MCP server?
Read-only browser MCP server — a local, configurable MCP shell around real Chrome for LLM agents.
What tools does browden expose?
2 in total: 2 read-only, 0 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.
Is browden safe to connect to an agent?
With care. The audit graded it B (87/100) and found 12 things worth knowing before you trust this server, listed below with the exact line each was found on.
What credentials does browden need?
No credential environment variables were found in its source, so it appears to need none.
How does browden run?
It speaks sse and stdio, so it runs as a local process your client starts. It is published on PyPI as browden.
How current is this page?
The grade is for one exact copy of the source (a9f16e72059f), read on 2026-10-11. The repository is watched and re-audited when it changes.