Crabeye Bridge FAQs

Question 1

How are tools discovered and named in Crabeye Bridge?

Accepted Answer

On startup, the bridge launches and connects to all configured upstream servers, discovering their tools. Each tool is then namespaced by its server name (e.g., `linear__create_issue`), indexed for fuzzy searching, and made available without naming collisions.

Question 2

What is Crabeye Bridge?

Accepted Answer

Crabeye Bridge is a developer tool that consolidates multiple AI (MCP) server connections into a single interface. It provides discovery, namespacing, and execution policies to streamline AI tool usage and optimize token consumption for AI assistants.

Question 3

Does Crabeye Bridge support secure credential management?

Accepted Answer

Yes, Crabeye Bridge includes an encrypted credential store for securely managing API keys and other secrets required by your upstream MCP servers, supporting per-server rate limiting and protecting sensitive information.

Question 4

What are execution policies and how do they work?

Accepted Answer

Execution policies allow you to define how tools can be used. You can apply per-tool and global policies (always, prompt, never) to control which tools run automatically, which require AI assistant approval, or which are completely blocked, giving you fine-grained control over AI interactions.

Question 5

How does Crabeye Bridge optimize AI assistant token usage?

Accepted Answer

Instead of injecting the schemas for every upstream tool into the AI assistant's context window, Crabeye Bridge exposes only two meta-tools (`search_tools` and `run_tool`). This significantly reduces context bloat, saving tokens, as tool schemas are only provided when actively searched for or run.

Crabeye Bridge

Crabeye Bridge

主要功能

使用案例

主要功能

使用案例