Self-hosted AI agent harness that builds, tests, and reuses its own tools

Open source, single Go binary. Runs on your own machine, handles multi-step work, searches local files,
schedules recurring tasks, and shares its sandboxed tool library with Claude Code, Codex,
and any other MCP client.

English · 繁體中文

Why Agenvoy

• Self-extending tools — Builds the missing tool instead of stopping.
• One sandboxed tool library — Shares the same tool library across Agenvoy, Claude Code, Codex, and other agents.
• Live command feedback — Streams throttled command output to the TUI and Web dashboard while preserving the complete final result.
• Skill-aware model routing — Passes matched Skill descriptions into agent selection so the dispatcher can choose models against the actual task contract.
• Audio model routing — Select separate speech-to-text and text-to-speech models from the configured OpenAI or Gemini providers.
• Origin-aware confirmations — Routes approval prompts back to the originating CLI, Web, Telegram, or Discord channel without cross-channel interception.
• Extensible integrations — Connects stdio or HTTP MCP servers, supports OAuth login, and refreshes remote tool catalogs.
• Self-hosted execution — Runs scheduling, memory, and file search on your own machine.
• Local-first agent — Runs on your personal computer, where the agent can safely work with your files, tools, memory, and schedules without moving them to a hosted service.
• Private chatbot access — Telegram and Discord connect back to the local daemon with a token, so users do not need to expose their host or configure inbound networking.
• Agent app and MCP server — Provides both roles from the same binary.

What you can do with it

Ask for live answers What's the weather in Taipei? The agent finds current data, calls tools, and gives you the answer. If a tool doesn't exist, it builds one.	Turn one sentence into automation Report TSMC stock price every morning at 8am The agent asks: Where to push results What format you want When to run Then creates the schedule automatically.
Ask your local files questions Find all invoices from last year Which document mentions Prompt guide? The agent searches your local files and answers directly.	Finish multi-step work Summarize today's GitHub commits and generate a progress report The agent breaks down the task, calls tools, combines results, and replies.

Works with the agents you already use

Agenvoy is also an MCP server.

Claude Code, Codex, OpenCode, and other AI agents can connect and:

• Use all your sandboxed tools
• Auto-build new tools when none exist
• Share every tool across all agents

One line of config. Instant shared tool library. Tools created in the demo: fetch_weather · fetch_crypto_price

Claude Code creates a weather tool (1)	Codex reuses it and creates a crypto tool (2)	Agenvoy tests both tools (3)

Who it's for

Agenvoy is for developers, technical operators, and AI-heavy workflows that need more than chat:

• People who want a self-hosted agent with sandbox guardrails
• Teams that want reusable tools across agents
• Users who need automation, file search, and scheduled reporting in one place

---

Core capabilities

Capability	Description
Auto tool generation	Builds and saves tools when they're missing
Self-scheduling	Create cron jobs with a single sentence
Long-term memory	Retains key info and context
Knowledge notes	Reads the notes you keep, before it answers
File search	Answers from your local files
Sub-Agent	Multi-agent collaboration
MCP client	Connect to external MCP services via official go-sdk (live tool refresh)
MCP server	Expose sandboxed tools to any MCP-compatible agent
Reasoning guides	On-demand rules via reasoning_guide(topic=...)
Tool Market	Share and install tools
Image generation	Generate images through a configured provider
Live command output	Stream run_command progress to the TUI and Web dashboard
Secure file boundary	Confirm sensitive paths and out-of-home access before granting them
MCP OAuth	Log in to HTTP MCP servers and persist tokens in the OS keychain
Transcription	Audio and video to text
Self-improvement	Auto-fixes after execution failures

---

Web Dashboard

Open web.agenvoy.com in your browser to reach the dashboard once the daemon is running on your machine.

▶ Watch the Web Dashboard walkthrough

Chatbot Integrations

Agenvoy currently supports Telegram and Discord as its chatbot channels. Both use outbound connections from the daemon, so your host does not need to be exposed to the public internet. Users only need to provide a bot token instead of configuring inbound ports, reverse proxies, or other network plumbing. Additional chatbot platforms will be considered only when they offer a meaningful security improvement.

Voice output is temporarily unavailable in Telegram and Discord; audio model settings remain available for local tool use and future channel support.

One-line install

On MacBook, also run sudo pmset -c sleep 0 to prevent sleep from interrupting schedules.

curl -fsSL https://agenvoy.com/scripts/install.sh | bash

---

Developer Recommendations

A cost-effective model setup to get started:

• Apply for a free NVIDIA NIM API token for gpt-oss-20b as the dispatcher model.
• Use a subscription plan as your primary model:
  • OpenAI ChatGPT Plus ($20/mo)
  • SuperGrok ($30/mo)

NVIDIA NIM no longer supports gpt-oss-120b, so it is no longer recommended as a fallback or summary model.

---

Docs

Full documentation at agenvoy.com/docs

• Getting Started
• Sessions & Agents
• Providers
• Built-in Tools
• MCP Server
• MCP Client
• Memory System
• Skill System
• Sandbox
• Architecture

License

This project is licensed under the Apache License 2.0.

Author

Just open an issue to share an idea.

---

©️ 2026 邱敬幃 Pardn Chiu