Agenvoy - An Agentic AI Framework

Name: Agenvoy - An Agentic AI Framework
Availability: InStock
Author: 邱敬幃（Pardn Chiu）

Go-based, zero third-party framework dependencies

by 邱敬幃（Pardn Chiu）

FreemacOSLinux

Visit website

Upvotes: 6
Comments: 0
Impressions: 263

About

Agenvoy is an open-source Go-based agent execution engine inspired by OpenClaw. It provides a unified interface over multiple LLM providers (including Copilot, Claude, OpenAI, Gemini, Nvidia, and OpenAI-compatible endpoints) and uses a planner model to route each request to an appropriate provider. Tasks are driven by declarative “skills” defined in Markdown that specify system prompts and permitted tools, and a selector chooses the best skill at runtime. Agenvoy includes a built-in toolset for filesystem operations, web access, scheduling, error memory, and HTTP requests, plus a JSON-driven extension system for adding external REST APIs without changing Go code. It emphasizes security through OS-native sandboxing for command execution, symlink-safe path validation with deny-lists, and storing API keys in the OS keychain. It can run as a CLI or as a persistent Discord bot with per-channel sessions and scheduled task callbacks, and it maintains cross-session memory via stored structured summaries and persisted tool errors.

Problem & solution

Problem

Hard to run agentic tasks across LLMs securely with tools, memory, and scheduling.

Solution

Go agent unifies LLM providers with skills, tools, sandboxing, memory, and bot/CLI modes.

Who it's for

Developers needing multi-LLM agent execution via CLI or Discord.

Main features

Multi-provider LLM routing
Sandboxed command execution
Skill-based tool allowlists
Token usage tracking
Persistent session memory