$ loading_
Access documents and many tools through an authenticated MCP gateway efficiently.
Copy the install command and let the AI configure it · recommended for beginners
No copy-paste install info for "Harbormaster" yet — see the docs or source repo.
First search the ingested documents for content related to “API gateway configuration”, then progressively discover available tools, choose the most relevant one, and explain why it should be invoked.
Returns relevant information from the documents and suggests the right next action through tool search, description, and invocation.
Based on my ingested documents, answer “What are the deployment dependencies for the current project?” and, if needed, progressively discover available capabilities through the tool registry.
Provides an answer grounded in that user’s documents and notes which discoverable tool capabilities were used if needed.
I need to run an automated investigation. First search for and describe available tools related to logs, configuration, and diagnostics, then choose the appropriate tool for the next step.
Lists relevant tool purposes and selects the most suitable tool from the registry to continue.
Researchers or developers can ingest documents and then query them through per-user RAG. This helps them get answers grounded in their own material.
When users do not know which tool to call, they can search, inspect descriptions, and then invoke the right one. It fits environments with many tools and progressive narrowing.
When documents and many tools must be accessed together, Harbormaster uses progressive discovery to keep context cost constant. It is useful for complex agent-style workflows.
It is an authenticated MCP gateway that ingests documents and orchestrates many tools through progressive discovery. It also provides per-user RAG over documents and a registry of 116 tools.
The description says it uses progressive discovery so the model can search, inspect, and then invoke tools while keeping context cost constant. For implementation details, see the source repository.
Yes. The provided information explicitly states that it offers per-user RAG over ingested documents for retrieval and question answering.
Unifies MCP clients with tool routing, memory, and automation flows.
Aggregate MCP servers and find tools through natural language semantic search.
Production-ready MCP server for query normalization, retrieval, and RAG prompt building.
Securely search internal documents and company data through an authenticated MCP connector.
Securely expose OpenAPI endpoints as MCP tools with auth, limits, and auditing.
Use one interface for search, code execution, files, and database queries.