
Connapse
Connapse gives AI agents persistent memory across sessions. It's a self-hosted knowledge backend that ingests documents from S3, Azure Blob Storage, local filesystems, or direct uploads, then exposes them through 11 MCP tools including semantic and hybrid search, bulk file operations, and container management. The agent can query your research corpus, upload new documents, and retrieve full parsed text with source citations. Everything runs in Docker with PostgreSQL and pgvector for storage, MinIO for objects, and a .NET 10 backend. Containers synced from cloud storage are read-only, while filesystem containers support writes based on permission flags. You'd reach for this when you need Claude to remember documents and research between conversations without re-uploading context every time.
- GitHub stars
- 13
- tools
- 3
- transport
- HTTP
- last updated
- Mar 14, 2026
Tools
3 toolsPublic tool metadata for what this MCP can expose to an agent.
search_knowledgeConnect an agent: Agents get 21 named tools, such as search_knowledge, upload_file, get_document, the code_* tools and the memory_* tools, plus 3 generic tools that reach the rest of the API.
Connect an agent: Agents get 21 named tools, such as search_knowledge, upload_file, get_document, the code_* tools and the memory_* tools, plus 3 generic tools that reach the rest of the API.
No parameter schema in public metadata yet.
upload_file
No parameter schema in public metadata yet.
get_document
No parameter schema in public metadata yet.
Sponsors
Readme
Connapse logo
The self-hosted knowledge base your AI agents search and remember with, over MCP
Your AI agents forget everything between sessions. Connapse fixes that.
Details
- Repository
- github.com/destrayon/connapse
- Categories
- Documents & Knowledge, Search & Web Crawling
- Package
ghcr.io/destrayon/connapse:v0.3.2- Transport
- HTTP
- Registry
- MCP registry, active