Obris

Stop re-explaining yourself to AI. Save knowledge once, use it in every conversation.

Obris is an MCP server that provides persistent knowledge storage for AI agents. Its core purpose is to remove the need to re-explain recurring context in every new conversation — save something once and it becomes available across all subsequent sessions. The server is distributed as the obris-mcp package on PyPI at version 0.4.0 and communicates over stdio transport. It requires an OBRIS_API_KEY environment variable to authenticate with the Obris service. The publisher is ai.obris and the source repository is hosted at github.com/obris-dev/obris-mcp. No tool list is exposed in the current registry record, so the specific read and write operations the server surfaces to an agent are not documented here. What the server is designed for is clear from its description: storing knowledge that would otherwise be repeated as manual context at the start of every chat — things like project conventions, personal preferences, domain glossaries, or standing instructions. An agent integrated with Obris can retrieve that stored knowledge automatically rather than waiting for a user to supply it again. This server is a poor fit for workflows that require structured relational data queries, real-time data pipelines, or any transport other than stdio.

Use cases

  • Store project-specific conventions once and have agents reference them automatically in future sessions
  • Persist personal preferences or standing instructions so they carry across separate agent conversations
  • Build a reusable domain glossary that agents can look up without being re-briefed each time
  • Reduce prompt length by offloading repeated background context to persistent storage

When to use it

  • When the same background context needs to be available to an agent across many separate sessions
  • When users or teams frequently re-supply the same information at the start of conversations
  • When building agents that should retain institutional or project knowledge between runs

When not to use it

  • When no OBRIS_API_KEY is available, as authentication is required
  • When the use case requires structured SQL or relational database queries
  • When the deployment environment cannot use stdio transport
  • When a full list of exposed tools is needed before adoption, as none are documented in the current registry record