Categories
self-hosted developer-tools mcp-server workflow-automation personal-assistant ai-assistant autonomous-agentsStars
Forks
Watchers
Developer links
OwnPilot
Bridging local hardware and cloud intelligence, OwnPilot orchestrates autonomous multi-channel agent swarms and visual workflow pipelines while keeping sensitive personal records strictly on private infrastructure. Users can trigger autonomous task runs that browse websites using headless browser automation, extract data tables, and summarize research directly into encrypted markdown notebooks. The built-in workflow canvas allows individuals to construct branching automation sequences combining large language models, conditional logic gates, iterative data loops, and human-in-the-loop approvals. Users delegate calendar scheduling, task tracking, expense budgeting, and contact lookups through conversational chats conducted over web portals, Telegram bots, or WhatsApp numbers. Operators can dispatch coding agents to inspect local repositories, execute shell commands in sandboxed environments, and stream live terminal logs as automated patches develop. The platform also enables home automation enthusiasts to monitor MQTT telemetric feeds from microcontrollers, queue device commands, and receive instant alert notifications when sensor thresholds trigger. Advanced security layers inspect all incoming prompts for personal data patterns, enforce strict permission boundaries across file directories, and generate tamper-evident audit logs for every autonomous tool call. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
Benefits
- Autonomous Multi-Channel Agent Workflow Automation
- Dispatches autonomous agents across Telegram, WhatsApp, and browser interfaces to execute multi-step plans, schedule calendar events, and manage personal data records with strict human-in-the-loop approvals.
- Universal Model Context Protocol Integration
- Acts as both an MCP client and server, allowing local agents to query external tools like GitHub and filesystem servers while exposing internal utilities to desktop Claude environments.
- Four-Layer Sandboxed Security and Privacy
- Protects sensitive data using AES-256-GCM encryption, automatic PII redaction across fifteen categories, granular permission matrices, and isolated execution sandboxes for untrusted code execution routines.
- Extensive Model and Tool Orchestration
- Routes requests dynamically across 104 AI providers including local Ollama models and cloud APIs, utilizing a meta-tool proxy that exposes 250 built-in utilities without bloating token context windows.
Features
- Visual Workflow Canvas
- Constructs automated execution graphs using twenty-four node types including LLM conditions, parallel branches, iterative loops, and webhook triggers with real-time copilot assistance.
- Headless Browser Automation
- Executes Playwright-powered Chromium browser sessions to navigate web pages, submit complex forms, capture visual screenshots, and extract unstructured DOM data via autonomous tool calls.
- MQTT Edge IoT Connectivity
- Integrates with Mosquitto MQTT brokers to register microcontrollers, ingest live sensor telemetry, queue bidirectional commands with acknowledgments, and trigger automated alerts.
- Unified Messaging Channel Gateways
- Connects conversational agents to Telegram bot APIs and WhatsApp accounts using Baileys socket libraries, enabling secure two-way messaging without requiring commercial enterprise subscriptions.
- Integrated Coding Agent Terminal
- Spawns and manages Claude Code, Codex, and Gemini CLI developer runtimes directly within the web dashboard with real-time pseudoterminal streaming and interactive command input.
- Multi-Provider Audio Pipeline
- Processes spoken voice inputs through Whisper speech recognition endpoints and synthesizes natural audio responses using OpenAI text-to-speech models across six configurable voice profiles.