Skip to content
Kitbay
MCP server

research-swarm

Runs locallyISCUpdated on Nov 10, 2026839GitHub starsNeeds an API key

Research Swarm is a command-line research tool by rUv, kept in the examples folder of the agentic-flow repository, that splits a research goal across three to seven agents running in parallel and plans sub-goals with goal-oriented planning, storing jobs in a local SQLite database.

What it does

Research Swarm is a command-line research tool by rUv, kept in the examples folder of the agentic-flow repository, that splits a research goal across three to seven agents running in parallel and plans sub-goals with goal-oriented planning, storing jobs in a local SQLite database. It can also run as an MCP server over stdio or HTTP, with six tools to set up the store, create, start and inspect research jobs and update their progress. It needs an Anthropic API key; Gemini, OpenRouter and Brave Search keys are optional, and a Supabase sync is available for shared setups. Building it compiles better-sqlite3, so Python and a C++ compiler are needed.

Capabilities

  • Create and start research jobs
  • Split a goal across parallel agents
  • Track job progress in SQLite
  • List and inspect past jobs