Skip to content

docs: state local MCP key requirements exactly - #251

Open
MagMueller wants to merge 1 commit into
browser-use:mainfrom
MagMueller:agency/mcp-key-contract-job1537
Open

docs: state local MCP key requirements exactly#251
MagMueller wants to merge 1 commit into
browser-use:mainfrom
MagMueller:agency/mcp-key-contract-job1537

Conversation

@MagMueller

@MagMueller MagMueller commented Sep 5, 2026

Copy link
Copy Markdown
Contributor

Clarifies that direct local MCP browser tools do not need an LLM key. Content extraction uses an OpenAI-compatible key. The autonomous agent tool uses that route by default and also supports configured AWS Bedrock credentials. Removes the unsupported Anthropic-key fallback claim.

Validated the absent-OpenAI/Anthropic-only error and the Bedrock routing branch without provider calls. The Bedrock constructor was replaced by a sentinel for routing only; AWS authentication and inference were not tested. Mint found no broken links, docs.json parsed, and git diff --check passed. The generated LLM-readable mirror was synchronized.


Summary by cubic

Clarifies the local MCP server's LLM key requirements so users don't set up unused keys. Direct browser automation tools like navigate, click, type, page state, HTML, and screenshot work without an LLM key; content extraction requires an OpenAI-compatible key, and the default autonomous agent route also uses OpenAI-compatible by default or AWS Bedrock when configured. Removes the unsupported Anthropic-key fallback from environment variables and troubleshooting.

Written for commit b068ec1. Summary will update on new commits.

Review in cubic

@cubic-dev-ai cubic-dev-ai Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

1 issue found across 2 files

Prompt for AI agents (unresolved issues)

Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="docs/open-source/customize/integrations/mcp-server.mdx">

<violation number="1" location="docs/open-source/customize/integrations/mcp-server.mdx:6">
P2: The intro advertises the Bedrock route (`MODEL_PROVIDER=bedrock` plus "configured AWS Bedrock credentials"), but the Environment Variables section on this page does not document `MODEL_PROVIDER` or the AWS credential variables (`AWS_ACCESS_KEY_ID`, `AWS_SECRET_ACCESS_KEY`, `AWS_DEFAULT_REGION`, as used in supported-models.mdx). A user following this page cannot configure the Bedrock route. Add the missing variables to the Environment Variables list.</violation>
</file>

Reply with feedback, questions, or to request a fix.

Fix all with cubic | Re-trigger cubic

---

Browser Use can run as a local **Model Context Protocol (MCP)** server on your machine via stdio. This is the **free, open-source option** that gives you direct, low-level control over browser automation but requires your own LLM API keys.
Browser Use can run as a local **Model Context Protocol (MCP)** server on your machine via stdio. This is the **free, open-source option** for direct browser control. Navigation, clicks, typing, page state, HTML, and screenshots do not need an LLM API key. Content extraction requires an OpenAI-compatible key. `retry_with_browser_use_agent` uses an OpenAI-compatible key by default or configured AWS Bedrock credentials when `MODEL_PROVIDER=bedrock`.

@cubic-dev-ai cubic-dev-ai Bot Sep 5, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2: The intro advertises the Bedrock route (MODEL_PROVIDER=bedrock plus "configured AWS Bedrock credentials"), but the Environment Variables section on this page does not document MODEL_PROVIDER or the AWS credential variables (AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_DEFAULT_REGION, as used in supported-models.mdx). A user following this page cannot configure the Bedrock route. Add the missing variables to the Environment Variables list.

Prompt for AI agents
Check if this issue is valid — if so, understand the root cause and fix it. At docs/open-source/customize/integrations/mcp-server.mdx, line 6:

<comment>The intro advertises the Bedrock route (`MODEL_PROVIDER=bedrock` plus "configured AWS Bedrock credentials"), but the Environment Variables section on this page does not document `MODEL_PROVIDER` or the AWS credential variables (`AWS_ACCESS_KEY_ID`, `AWS_SECRET_ACCESS_KEY`, `AWS_DEFAULT_REGION`, as used in supported-models.mdx). A user following this page cannot configure the Bedrock route. Add the missing variables to the Environment Variables list.</comment>

<file context>
@@ -3,7 +3,7 @@ title: "MCP Server"
 ---
 
-Browser Use can run as a local **Model Context Protocol (MCP)** server on your machine via stdio. This is the **free, open-source option** that gives you direct, low-level control over browser automation but requires your own LLM API keys.
+Browser Use can run as a local **Model Context Protocol (MCP)** server on your machine via stdio. This is the **free, open-source option** for direct browser control. Navigation, clicks, typing, page state, HTML, and screenshots do not need an LLM API key. Content extraction requires an OpenAI-compatible key. `retry_with_browser_use_agent` uses an OpenAI-compatible key by default or configured AWS Bedrock credentials when `MODEL_PROVIDER=bedrock`.
 
 <Note>
</file context>
Fix with cubic

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant