Quilombo

Quilombo is agent-first storage for physical inventories: workshops, libraries, tool rooms, and anywhere people need to put things away and find them again.

The storage layer is deliberately conservative. Agents and clients interpret observations, ask for confirmation, and decide when a fact is uncertain. Quilombo stores locations, items, holdings, spatial relations, provenance metadata, and workspace-scoped permissions.

        flowchart LR
    A[Person] --> B[Agent or client]
    B -->|MCP or REST| C[Quilombo]
    C --> D[(Workspace inventory)]
    B -->|interprets photos,
    language, intent| B
    

Learn Quilombo

• Getting started
  • Run locally
  • Create a workspace token
  • Record inventory
• Concepts
  • Workspaces
  • Locations
  • Items and holdings
  • Search
  • Labels
  • Provenance
  • History and undo
  • Freshness
• Labels and emerging facets
  • Vocabulary
  • Deterministic identity
  • Write and suggestion flow
  • Workshop and library examples
  • Merge and facet design
• Audits and freshness
  • Freshness states
  • Re-verify while already there
  • MCP audit

Integrate

• MCP integration
  • Tools
  • Error responses
  • ChatGPT and Claude
• Import and export
  • Export
  • Import
• Agent Plugin
  • Package layout
  • Versioning and ownership
• Plugin evaluation
  • Positive cases
  • Negative cases
  • Evidence

Architecture

• Architecture

Contribute

• Development
  • Environment
  • Local PR testing with production-shaped data
  • Releases
  • Build the documentation
  • Project workflow

The generated REST schema is available from a running installation at /api/docs/.