# paradigmxyz/centaur

Centaur is frontier, agentic infrastructure that you own. Centaur is like Claude Tag, but open source and on steroids.

Repository: https://github.com/paradigmxyz/centaur
Canonical: https://ross.abutalabs.com/products/centaur
Homepage: https://centaur.run
Language: Python
License: NOASSERTION
License Family: other
Last push: 2026-09-02T19:02:16+00:00

## Health v2 (maintenance only)
Score: 77/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 100, release rhythm 88, longevity 7
- inputs: {"age_days": 107, "days_push": 0, "days_rel": 3, "gap_med": 0, "n_releases_24m": 72}
- flags: young, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1218, forks 228 (observed 2026-09-03T02:15:16.719327+00:00)

## What it is
Centaur is a self-hosted, open-source platform for running shared AI agents for teams, triggered via Slack mentions or an API. Each conversation runs in an isolated Kubernetes sandbox with real dev tools, durable replayable state, shared Python tool plugins, and credential-safe egress so agents never see raw API keys.

## Use cases
- run a shared AI agent for my team in Slack
- let an agent investigate CI failures in a real repo
- self-host an open-source alternative to Claude Code for teams
- give agents access to internal tools without exposing API keys
- run durable multi-step agent workflows that survive restarts
- add custom Python tools to agent conversations
- summarize Slack threads and customer context with an agent

## When to choose
- your team wants one shared Slack-native agent instead of individual local setups
- you need sandboxed, real execution environments with shell, git, and code runtimes
- credential safety matters - agents must call services without holding raw keys
- you want durable, replayable agent runs and workflow orchestration
- you want to bring your own agent harness like Claude Code, Codex, or Amp

## When to avoid
- you only need a simple chatbot with no code execution or tool use
- you cannot run Kubernetes or a k3s cluster in your infrastructure
- you want a fully managed SaaS with zero infrastructure to operate
- your use case is a single-user local coding assistant rather than a team platform

## Facets
- artifact type: application
- maturity: active
- function: agent-framework, chatbot, workflow-automation, secrets-management, self-hosted, sdk, plugin-system
- domain: large-language-models, self-hosted, developer-tools, messaging-platforms
- platform: python, self-hosted, cli
- tags: slack-integration, agentic-infrastructure, sandboxed-execution, durable-workflows, credential-proxy, team-agents, harness-agnostic, ai-agents, automation, kubernetes, docker, web-server

## Member repositories
- paradigmxyz/centaur (main) score 77

## Provenance
- Observed fields: from GitHub, fetched 2026-09-03T02:15:16.719327+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:23:19.897915+00:00, confidence not recorded.
  - readme: https://github.com/paradigmxyz/centaur (fetched 2026-09-03T02:15:16.719327+00:00, sha 35e0956a7124)
  - homepage: https://centaur.run (fetched 2026-08-29T12:30:48.477390+00:00, sha 0391224aa278)
  - site_page: https://centaur.run/architecture (fetched 2026-08-29T12:30:48.486322+00:00, sha dd6e91e9515a)
- Data as of 2026-08-30T08:39:29.467469+00:00.