Loading ToolBench

Reading the current corpus…

mcp-server-browserbase MCP server · F · ToolBench

mcp-server-browserbase

github.com/Matvey-Kuk/mcp-server-browserbase

Cloud browser automation server powered by Browserbase and Stagehand. Enables LLMs to navigate websites, interact with elements, extract data, and capture screenshots using natural language commands.

<=2025-11-25STDIO16 tools

47/100

Definition Quality46

Protocol Readiness47

Supportability50

Spec posture

Inferred effective spec <=2025-11-25·current baseline 2026-07-28

Static source inference · medium confidence · detected: Sampling

Deprecated protocol patterns detected

Summary

Server has 16 tools with basic schemas and descriptions, but significant gaps limit production readiness. All tools have descriptions (10-50 chars, mostly short), and input schemas are present with type definitions. However, descriptions are generic and lack LLM-optimized guidance on WHEN to use each tool or WHAT happens (especially for write operations). Parameter descriptions are minimal. Output schemas are not documented. Error handling is absent from visible code. Multi-session tools duplicate single-session tools without clear differentiation. No tool annotations (readOnlyHint/destructiveHint) despite clear risk classifications.

Tools (16)

browserbase_session_createwriteauth50/100

Create a new browser automation session

browserbase_stagehand_actwriteauth50/100

Perform an action in the browser using natural language

browserbase_stagehand_extractread onlyauth50/100

Extract structured data from the browser using natural language

browserbase_stagehand_get_urlread onlyauth50/100

Get the current URL from the browser

browserbase_stagehand_navigatewriteauth50/100

Navigate to a URL in the browser

browserbase_stagehand_observeread onlyauth50/100

Observe and analyze the current state of the browser

browserbase_stagehand_screenshotread onlyauth50/100

Capture a screenshot of the current browser state

multi_browserbase_stagehand_act_sessionwrite

Issues (5)

HIGHtool-description ↗

Descriptions are too short (10-50 chars) and lack LLM-optimized guidance. Missing WHEN to use each tool, WHAT it returns, and dependencies. E.g., 'Perform an action in a specific multi-session browser using natural language' does not explain what 'action' means or what the tool returns.

multi_browserbase_stagehand_act_sessionmulti_browserbase_stagehand_extract_sessionmulti_browserbase_stagehand_observe_sessionbrowserbase_stagehand_actbrowserbase_stagehand_extractbrowserbase_stagehand_observe

HIGHresponse-shaper ↗

No output schemas documented. LLMs cannot plan downstream calls or extract data without knowing what fields to expect. E.g., does extract_session return JSON, text, or structured data?

multi_browserbase_stagehand_extract_sessionbrowserbase_stagehand_extractmulti_browserbase_stagehand_observe_sessionbrowserbase_stagehand_observe

HIGHtool ↗

Recommendations

Expand tool descriptions to 50-200 chars following LLM-optimized pattern: 'WHAT does it do? WHEN to use it? WHAT does it return?' E.g., 'Extract structured data from the current browser page using natural language instructions. Returns JSON object with extracted fields. Use after navigate() to parse page content.'
Document output schemas for all tools. For extract_session and observe_session, specify: 'Returns {success: boolean, data: object, error?: string}' or similar. This enables LLMs to chain calls correctly.
Consolidate single-session and multi-session tools. Either: (1) merge into one tool with optional sessionId parameter (defaults to implicit session), or (2) rename multi-session tools to clarify intent, e.g., 'browserbase_stagehand_act_in_session' vs 'browserbase_stagehand_act'.
Add tool annotations to all tools. Mark WRITE operations with destructiveHint=true, READ operations with readOnlyHint=true, and idempotent operations with idempotentHint=true. This guides agent retry and error recovery logic.
Enhance parameter descriptions with format, constraints, and examples. E.g., 'sessionId: UUID of the browser session (e.g., 'sess_abc123'). Obtain from session_create() or session_list().' and 'action: Natural language instruction (e.g., "Click the login button and enter username"). Max 500 chars.'

Spec posture evidence

Inferred effective spec: <=2025-11-25.

Relies on Sampling (deprecated) - integrate directly with the LLM provider API

Score history

Overall score trend

First recorded score · v2 rubric

47/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-23	F	47	<=2025-11-25	v2