Open Source
Local-First AI Agent
Rust-powered AI that works on your machine.
Coding, research, background agents, and automation.
An AI agent that belongs on your desktop. An embedded web UI, local storage, and no Node, Python, or Docker required for the core app. Bring a cloud provider or run your models locally.
Core Capabilities
Everything you need to run AI agents on your own infrastructure.
200+ Providers
A bundled catalog of 200+ providers and 7,700+ model entries. Connect to cloud services or local Ollama models; availability depends on your provider and account.
Web UI
Workspaces, streaming tool cards, working notes, context compaction, and searchable conversation history.
40+ Built-in Tools
File edits, bash, live code search, LSP diagnostics, web research, memory, and planning. Load less-used tools on demand.
Voice I/O
Optional local Whisper transcription and Piper speech, with browser fallback. Local voice needs additional binaries and models.
Skills System
Install .oskill bundles or let OSA create and update skills at runtime. Connect MCP servers for additional tools.
OAuth Login
Sign in with GitHub Copilot, Google, or OpenAI Codex. No API keys needed.
Visual Workflow Editor
Node-based drag-and-drop automation. Build complex pipelines visually with conditions, loops, and branching logic.
Discord Bot
Optional Discord integration with channel sessions, community access controls, audio transcription, and tool progress. Music playback is available in voice-enabled builds.
Jobs Scheduling
Schedule reminders, recurring tasks, and daily briefings with cron expressions or natural language.
Built for longer-running work
Delegate tasks, keep context, and extend your assistant as you go.
Background Subagents
Delegate parallel tasks to resumable subagents. Completed work returns to the parent conversation automatically.
Context That Carries Forward
Rolling working notes and context compaction support longer sessions. Search past conversations when you need to revisit a decision.
In-App Updates
Stay current with in-app updates, checksum-verified downloads, and resumable transfers. See the changelog for recent improvements.
Benchmarks
Measured with in-repo runtime benchmarks (debug vs release, provider-free workloads).
Release avg over 10 runs to first successful /api/auth/status (p50 ~513ms)
Resident memory sampled immediately after server readiness
Resident memory after 2s idle settle window
Latest checked-in report: 2026-08-31, 10 runs per profile. Reproduce: cargo run --release --bin osagent-bench -- --profiles debug,release --iterations 10. Server readiness is not full UI readiness or model inference. Results vary by machine, build, and workload.
"Making it feel like a useful assistant integrated into your daily routine. Not just another tool you open occasionally, but something that actually helps with real tasks."
OSAgent Design Philosophy
Whether it's coding, research, automations, or just having a conversation while you're working. The dream is an agent that feels like it belongs on your desktop.
Architecture
Built with Rust for performance and reliability. Runs at http://localhost:8765 with optional GUI launcher. Read our research on self-optimizing prompts for more.
Super-duper fast
OSAgent keeps the system compact: Rust on the backend, a static web UI on the frontend, local persistence by default, and optional subsystems only when you need them.
Runtime
Tokio
Port
localhost:8765
Config
~/.osagent/
License
MIT
Integrations
Connect to your existing tools and platforms.
Frequently Asked Questions
Common questions about OSAgent, local-first AI agents, and self-hosting.
What is OSAgent?
OSAgent is a local-first AI agent built with Rust for coding, research, and automation. It connects to cloud providers or local models, with an embedded web UI and optional Discord and voice interfaces.
How is this different from Cursor or Windsurf?
OSAgent is a general assistant rather than a dedicated editor. Coding tools sit alongside web research, scheduled jobs, persistent memory, and optional Discord and voice interfaces. The Rust core embeds its web UI and stores sessions locally. Cloud model requests go to the provider you configure; local inference is available through Ollama.
What do I need to get started?
Install OSAgent and open http://localhost:8765. Connect a provider using OAuth or an API key, or use local Ollama models. The core app needs no Node, Python, or Docker; optional tools and voice features may need additional runtimes or model downloads.
Can I run it fully offline?
Yes, for local model and workspace tasks: install Ollama and download a model first, then configure OSAgent to use it. Cloud providers, web search, Discord, and update downloads still require a network connection.
What models do you recommend?
Choose a model with reliable tool calling for coding and automation. Use a stronger reasoning model for planning and debugging, a faster model for routine tasks, or a local model that fits your hardware for privacy. Compare cost, context limits, and results on your own tasks rather than relying on one permanent ranking.
How does Discord integration work?
Use a Discord-enabled build, configure your bot token and access rules, and run it alongside the web UI. It supports channel sessions, slash commands, community access controls, and audio transcription. Music playback requires the optional discord-voice build feature.
What's the goal for OSAgent?
Making it feel like a useful assistant integrated into your daily routine. Not just another tool you open occasionally, but something that actually helps with real tasks, whether that's coding, research, automations, or just having a conversation while you're working. The dream is an agent that feels like it belongs on your desktop, not in the cloud.
How much does it cost?
OSAgent is free and open source under MIT. Cloud providers may charge for API usage or subscriptions. Local inference avoids API fees, but still uses your hardware and electricity.