Self-host Trinity
v1.0.0Self-hosted AI agent platform for Claude Code, Codex, Gemini
curl -fsSL https://raw.githubusercontent.com/abilityai/trinity/main/install.sh | bash- Stars
- 711
- RAM
- 8 GB minimum (16 GB recommended for multiple agents)
- Ports
- 80 (Frontend web UI (HTTP)), 8000 (Backend API), 8080 (MCP server), 8686 (Vector log aggregation API), 4317 (OpenTelemetry gRPC receiver), 4318 (OpenTelemetry HTTP receiver), 8889 (Prometheus metrics exporter), 13133 (OpenTelemetry health check)
- Requires
- Redis, Vector
ghcr.io/abilityai/trinity-agent-baseWhat is Trinity?
Trinity is an Apache 2.0 licensed platform for running AI agents in production on your own hardware. It supports three execution environments — Claude Code, OpenAI Codex, and Gemini CLI — each running in its own isolated container. Agents are grouped into departments that reflect your organizational chart, complete with roles, measurable objectives, and reporting hierarchies. Every agent maintains its memory as a git repository you control, and all tool invocations are recorded in an immutable audit trail with an optional cryptographic hash chain. Actions that cannot be undone — such as sending funds, transmitting messages, or deleting data — are blocked until a designated human approves them through an Inbox interface. The built-in Workspace offers threaded chats, a live task card, a shared canvas, drag-and-drop file sharing, voice conversations, and the ability to rate deliverables. Agents can also be reached through messaging apps, voice calls, MCP, REST endpoints, and a command-line interface. Deployment options include a single-command Docker Compose installer, one-click images on DigitalOcean and AWS Marketplace, or manual setup on any server with Docker. The control plane consists of a Vue frontend, FastAPI backend, MCP server, scheduler, Redis for secrets and locking, and PostgreSQL (with SQLite as the default). The Enterprise tier adds OIDC and SAML single sign-on, two-factor authentication, role-based access control, a credential vault, compliance reporting, and cross-agent Projects. The vendor, Ability AI, operates their own business on Trinity with dozens of agents.
Should you self-host Trinity?
Trinity scores 57/100 on selfhost.directory (how scoring works) and ranks #60 of 209 in AI Agents. It has 711 GitHub stars.
It shipped 4 stable releases in the past year. The latest, v1.0.0, came out on 20 October 2026. New versions land roughly every 2 weeks.
Setup is advanced: plan for several services and some ongoing maintenance. An official Docker image is available. It is Apache-2.0-licensed and written mostly in Python.
Its closest alternative on selfhost.directory is Orca (91/100, 108 739 stars). It scores higher overall, so compare the two before you pick. See Trinity vs Orca side by side.
Features
- Cornelius cognitive architecture
- Credential vault (Enterprise)Enterprise only
- Department and role organization
- Git-native agent state
- Guardrails and cost controls
- Human Approval Gates
- MCP Server
- Multi-channel access
- Multi-runtime agent execution
- Projects (Enterprise)Enterprise only
- Shared Workspace
- Skills library and marketplace
- SSO and 2FA (Enterprise)Enterprise only
- System agent
- Tamper-evident audit log
Trinity README
More in AI Automation
trypostSelf-hosted social media scheduler with MCP for AI agents — publish to 13 networks, visual calendar, approvals, analytics, all features included.
VisionClaw-Agent-Public-ReleaseSelf-hosted AI operations platform where specialist agents collaborate through hundreds of tools to deliver real business assets on your infrastructure.
z8runSelf-hosted visual flow engine for automation and AI pipelines. Rust + React, drag-and-drop editor, 39 nodes, 12 AI nodes, WASM plugins, single binary with SQLite.
ZimaOS-BlueZimaOS Blue is an open-source, self-hosted AI agent runtime in Go that runs locally on any device with voice, browser automation, deep research, and 20+ IM integrations.
5dive5dive runs persistent AI agents as systemd services on your Linux server with a shared SQLite task queue and human-in-the-loop approval via Telegram/Discord.
aci-devACI.dev is an open-source tool-calling platform you self-host as a backend and dev portal, wiring 600+ SaaS integrations into AI agents via function calls or unified MCP.
action-syncSelf-hosted open-source tool using Claude AI to analyze meetings, search GitLab codebases, and create detailed Jira tickets automatically.
agentgatewayOpen-source, self-hosted HTTP and gRPC gateway routing LLM, MCP, A2A and service traffic in one binary, with auth, policy, guardrails, cost controls and observability.
agentic-chatopsSelf-hosted agentic ChatOps platform with 3-tier AI architecture that triages alerts, predicts remediation impact via causal graph, and auto-resolves routine issues.
Agentic SignalSelf-hosted visual workflow automation with local AI via Ollama. Drag-and-drop nodes for data, APIs, RAG — no cloud dependencies.
agentsInkeep Agents lets teams build AI assistants and workflows via drag-and-drop builder or TypeScript code with full sync. Self-host via Docker or Vercel.
AgenvoySelf-hosted AI agent running locally, auto-generates tools, shares via MCP with Claude Code and Codex, keeps data private.










0 comments