Deploy your own
AI agent fleet.
Spindrel is a self-hosted platform for deploying AI bots that manage your projects, review your code, monitor your systems, and automate your workflows.
Your entire RAG loop, silk-wrapped.
# Clone and run
git clone https://github.com/spindrel-dev/spindrel.git
cd spindrel
./setup.sh
# Your fleet is live at http://localhost:8000An AI operations platform
you actually own.
A code review bot that comments on every PR. A research assistant that builds knowledge over time. A project management hub with AI-managed task boards. A DevOps monitor that alerts you on Slack. Spindrel runs on your hardware with your choice of LLM — each bot gets its own personality, tools, and domain expertise, orchestrated through persistent channels with real memory.
Any LLM Provider
OpenAI, Anthropic, Gemini, Ollama, OpenRouter, LiteLLM, vLLM — mix and match providers per bot.
Self-Hosted
Your data stays on your hardware. No cloud dependencies, no data leaving your network.
Composable
Layer skills, carapaces, tools, and integrations onto any bot. Build once, reuse everywhere.
Built for orchestration at every layer.
Everything you need for autonomous AI agents.
Multi-Agent Fleet
Deploy specialized bots — each with its own model, persona, tools, and expertise. Bots delegate tasks to each other automatically.
Perfect for engineering teams with reviewers, QA, and leads.Carapaces
Composable expertise bundles — snap skills, tools, and behavioral instructions onto any bot at runtime.
Give any bot code-review, research, or PM expertise in one toggle. Explore Skills Hub →Persistent Memory
File-based memory: MEMORY.md, daily logs, and reference docs. All on disk, all indexed for RAG, nothing opaque.
Research assistants that remember findings across sessions.Tasks & Scheduling
Heartbeats, recurring tasks, one-off deferred runs, cross-bot delegation — all managed by background workers.
DevOps monitors that check systems every hour automatically.Tool System
Local Python tools, MCP servers, and client-side actions. Tool RAG selects only relevant tools per query.
Bots that create slides, manage tasks, or query databases.Workflows
Multi-step automations with conditions, approval gates, and cross-bot coordination. Define in YAML, trigger from anywhere.
Content pipelines with draft, review, and approval steps.Channel Workspaces
Per-channel file stores with 17 built-in templates or create your own. Active files auto-injected, archives searchable, everything indexed.
Each project gets its own organized file store and context.Integrations
Slack, GitHub, Discord, Gmail, Frigate, ARR stack, and more. Build custom integrations with the plugin framework.
Connect to the services your team already uses. Browse Integrations →Docker Sandboxes
Isolated code execution in long-lived containers. Per-bot profiles, admin locking, configurable resources.
Bots that safely run and test code in isolation.Smart Retrieval
Hybrid BM25 + vector search, contextual retrieval, cross-encoder reranking — all with local models, zero API cost.
Large knowledge bases where dumping everything doesn't scale.
Works with any LLM provider. Mix and match per bot.
Ready to deploy your fleet?
Get started in minutes with Docker Compose.