mcp@4.8.1

VERIFIED

Publisher claims: "An MCP interface into the Bright Data toolset" Observed in sandbox: calls api.brightdata.com only, touches nothing outside the project, writes no memory.

Registry npm (MCP server)Publisher brightdata · publishing for 593 daysInstalls 69,303Last detonated 12 min ago
$claude mcp add mcp -- npx -y @brightdata/mcp
WatchReport an issue

Coverage

Started with a decoy API_TOKEN; tool list observed. Tool calls with a real credential not tested.
Tested
server startup with a tagged decoy in API_TOKEN · MCP initialize and tools/list · every tool description, for hidden instructions · where the decoy credential went · every file read, file written and host contacted by what ran
Not tested
what the tools do when called with a real credential
Needs
API_TOKEN — give it to the server where it runs, through vet-skills run --secrets, so the value never enters the server's environment. Never paste a credential into a chat, a ticket or a report.

Before you give it your token

It will ask for
API_TOKENConfiguration, not credentials: BASE_MAX_RETRIES, BASE_TIMEOUT, BROWSER_FILTERED_SNAPSHOT, BROWSER_ZONE, CONTENT_GATE, CONTENT_GATE_PROGRESS_MS, CONTENT_GATE_TIMEOUT, GROUPS, RATE_LIMIT, SERP_ZONE, TOOLS
Where it talks to
brightdata.com (its own)Also chatgpt.com, perplexity.ai — not the publisher's domain; check what for before connecting.
What we saw when it ran
no credential files read · reached only api.brightdata.com (its own)
What we could not test
what the tools do when called with a real credential

Connect it with a scoped credential, under the guard, and read the third-party hosts first. The code names chatgpt.com, perplexity.ai as well as its own vendor. The guard injects API_TOKEN only into requests to brightdata.com (and its subdomains) and refuses the rest.

$vet-skills run --secrets API_TOKEN=brightdata.com -- npx -y @brightdata/mcp

Never paste a credential into a chat, a ticket or a report, including ours. How the guard holds it.

What changed since 4.8.0

vet-skills diff mcp 4.8.0 4.8.1

In the sandbox, this version reached for nothing 4.8.0 did not. Compared from the two recorded runs, not from source.

Only one of the two runs produced a tool list; tools are not compared.

The guard can hold an update like this until someone approves it (runtime.on_change: hold). A change that only shows with a real credential, or on a path the run did not reach, is not in a trace and so not in this comparison.

Is [email protected] safe to install?

KillerSkills ran it in the KillerSkills honeypot sandbox (2 scenarios) on 2026-11-22 and scored it 89/100 (Verified). It did what it claims and nothing else we could detect: no undeclared network egress, no credential reads, no persistent instruction writes. Summary: calls api.brightdata.com only, touches nothing outside the project, writes no memory.

Behavior trace

Honeypot run · 52 s · 2 scenarios · 0 deviations
t+CategoryObserved action
0.0sprocessno install scripts in bundle
12sprocessdependency install exit 0
32sprocessrelaunched with a decoy API_TOKEN (tagged; nothing real)
52snetworkCONNECT tcp://api.brightdata.com:443
52sclaimed30 tools with a decoy API_TOKEN: ask_brightdata_assistant, search_engine, scrape_as_markdown, search_engine_batch, scrape_batch, list_dataset_fields, search_dataset, search_scrapers, get_scraper_details, run_scraper, get_scraper_progress, get_scraper_results, refresh_scrapers, session_stats, chatgpt_ai_insights, perplexity_ai_insights, scraping_browser_navigate, scraping_browser_go_back, scraping_browser_go_forward, scraping_browser_snapshot, scraping_browser_click_ref, scraping_browser_type_ref, scraping_browser_screenshot, scraping_browser_network_requests, scraping_browser_wait_for_ref
52stools583f787d485a1854
52stools7db766a2ae96f74b
52stools593ace8f97ab43f6
52stools26c12123029792df
52stools23851d8f881fed10
52stools00ed85d9186c3973
52stoolsfe75bb85695a8a9c
52stools835a9b9e15cc52ba
52stools471bbe2b98a05dca
52stoolse507345702d9ff37
52stoolsa2ee02b3702512bb
52stoolsca8bb9a37e623c7c
52stools4a57ff15477d0e03
52stoolsd5358e3175613934
52stools35ae7722c68ed148
52stools68097f918d94a85a
52stoolsd1f18642040b1386
52stools5d6ac0288e9eb7f5
52stools9ee87d913990bbdc
52stools35c3f2c527dd74d9
52stools36f26e22f0185714
52stools611626bd5b319846
52stoolseaeaa221c33939bc
52stools7cd32e7676dee1b6
52stoolsa550cb51ffb2b66e
52stoolsf8e236844faefbb3
52stools27987632cf3cfbba
52stoolsa65858732770ec54
52stoolsb502a5c560ec826d
52stoolsbf8e775f4ebdfdb0
52sprocessscript backend runs setup only
52sprocessno install scripts to re-run

Static findings

HIGH
Collapsed <details> block carrying instructionsREADME.md:168 · Collapsed <details> block carrying instructions<summary><b>Zed</b></summary> Add to your Zed settings: ```json { "context_servers": { "brightdata": { "url": "https://mcp.brightdata.com/mcp?token=YOUR_API_TOKFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Collapsed <details> block carrying instructionsREADME.md:185 · Collapsed <details> block carrying instructions<summary><b>Warp</b></summary> Go to Settings > MCP Servers > Add MCP Server and add: ```json { "brightdata": { "url": "https://mcp.brightdata.com/mcp?token=YOUFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Collapsed <details> block carrying instructionsREADME.md:90 · Collapsed <details> block carrying instructions<summary><b>Claude Code</b></summary> ```bash claude mcp add --transport http brightdata "https://mcp.brightdata.com/mcp?token=YOUR_API_TOKEN" ```Fix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Collapsed <details> block carrying instructionsREADME.md:99 · Collapsed <details> block carrying instructions<summary><b>Cursor</b></summary> Add to `~/.cursor/mcp.json`: ```json { "mcpServers": { "brightdata": { "url": "https://mcp.brightdata.com/mcp?token=YOUR_API_TOFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Collapsed <details> block carrying instructionsREADME.md:116 · Collapsed <details> block carrying instructions<summary><b>VS Code</b></summary> Add to `.vscode/mcp.json`: ```json { "servers": { "brightdata": { "type": "http", "url": "https://mcp.brightdata.com/mcp?tokenFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Collapsed <details> block carrying instructionsREADME.md:64 · Collapsed <details> block carrying instructions<summary><b>Claude Desktop</b></summary> 1. Go to: Settings → Connectors → Add custom connector 2. Name: `Bright Data` 3. URL: `https://mcp.brightdata.com/mcp?tFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Collapsed <details> block carrying instructionsREADME.md:134 · Collapsed <details> block carrying instructions<summary><b>Windsurf</b></summary> Add to `~/.codeium/windsurf/mcp_config.json`: ```json { "mcpServers": { "brightdata": { "serverUrl": "https://mcp.brightdata.Fix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Collapsed <details> block carrying instructionsREADME.md:151 · Collapsed <details> block carrying instructions<summary><b>Gemini CLI</b></summary> Add to `~/.gemini/settings.json`: ```json { "mcpServers": { "brightdata": { "httpUrl": "https://mcp.brightdata.com/mcp?tokeFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01

Cost to the agent

Instructions loaded
≈46 tok
1 file · 22 in bundle
Tool list per prompt
≈5,143 tok
30 tools · 171 per tool
Handshake
2513 ms
launch to tools/list, stdio
Undocumented tools
0
description under 20 chars · avg 226

Tokens estimated at four characters each. The tool list is re-read by the agent on every turn, so it is the number to compare between servers that do the same job: ranked across the index.

Known vulnerabilities

Not checked against OSV yet; the weekly check has not reached this version.

mcp@4.8.1 safety verdict: Verified — KillerSkills