EvoForge

by smallchaoVerified

一个高性能的具备进化能力的桌面智能体工作台

103
Stars
0
Forks
8/24/2026
Added
View on GitHubDownload ZIP

⚠️ Third-Party Software Notice

This skill is third-party open-source software developed and hosted independently on GitHub. SkillTip is an informational directory and does not control or maintain the underlying repository. Any security checks displayed are automated and limited in scope. Review the source code before installing.

Read the Terms of Service

Installation

Add to your Claude Code skills directory:

# Add to your Claude Code skills
git clone https://github.com/smallchao/EvoForge

Getting Started

Guides for using skills like EvoForge.

Security Report

Verified

Last scanned: —

{
  "status": "PASSED",
  "issues": []
}

README.md

EvoForge

Most AI agents start every day from zero.
EvoForge remembers everything it learned yesterday.

Version 0.4.2 Windows and macOS Built with Rust and Tauri Local first Six UI languages

Download · Why it's different · A real workbench · Connections · Security & privacy · Feedback


This repository hosts EvoForge releases, changelogs, and issue tracking. EvoForge is not an open-source project; nothing in this repository grants an open-source license.

⚡ Why it's different

Every agent you've tried has the same flaws: teach it something today, and tomorrow it's gone. It can run commands, but files, terminal, browser, and Git live in five different places. And you don't dare leave the screen during a long task, because you don't know what it might delete.

EvoForge's answer is a loop, not a chat box:

flowchart LR
    A([Goal]) --> B([Plan])
    B --> C([Execute & Verify<br/><small>browser · terminal · files · git</small>])
    C --> D([Deliver])
    D --> E([Distill<br/><small>skills · memory</small>])
    E -. next time: faster, sharper .-> C

After every task, EvoForge reflects on the run: action sequences worth repeating are distilled into skills, your preferences and project context are written into memory, and a built-in curator periodically merges, dedupes, and prunes what accumulates — so it doesn't bloat with use, it sharpens. The next time a similar task shows up, it reaches for the skill it already earned and skips every pit it already fell into.

$ "ship the landing page"

  ▸ plan       3 steps, 2 tools required
  ✓ browse     collected 14 sources
  ✓ terminal   build passed, 42 tests green
  ✓ files      wrote src/landing/…  (+412 −0)
  ✦ distilled  new skill: deploy-preview-flow
  ● memory     noted: you prefer pnpm

🛠 A real workbench

Not "a chat box with tools" — everything a worker needs, in one window:

🌐 BrowserEmbedded browser panel — the agent scrapes, clicks, and live-casts the page back to you
💻 TerminalReal shell execution; on Windows, encoding and error streams are handled automatically, output stays clean
📁 Files & GitRead/write the workspace, generate patches, commit branches, open GitHub PRs directly
🎯 Goal modeHand over a long-running goal; it plans, executes, and verifies autonomously within a token budget you set
🧠 Memory & skill panelsEverything it accumulates is visible, editable, deletable — you audit its "experience"
🔍 Full-text searchEvery past session is searchable; yesterday's conclusions don't need re-deriving

🔌 Connect your stack

The big three, one click each. GitHub, Cloudflare, and Supabase connect through their official CLIs with browser authorization — click, log in, done. More importantly, the agent knows what you've connected. With Supabase linked, ask it to "add an index to this table" and it just does it — no more "I don't have database access."

The long tail, through MCP. Any MCP server — stdio or remote HTTP — works by pasting its config. Servers that require OAuth (Notion, Linear, Sentry, Supabase MCP, and most official remote servers) connect with a single Authorize click: discovery, client registration, and PKCE are fully automatic; tokens persist and refresh on their own.

🎨 Looks matter too

StyleCharacter
Cold GrayThe Linear-inspired default — cool, precise, quiet
PaperWarm ink-on-paper, low glare, easy on long sessions
MistNordic blue-gray, calm and atmospheric
GraphitePure high-contrast monochrome, true-black dark mode for OLED

Four complete interface styles × five accent colors × light/dark = 40 combinations, every one calibrated for contrast, switchable live from Settings. And the UI speaks six languages: English (default), 简体中文, 日本語, 한국어, Deutsch, Français.

🔒 Security & privacy

Everything stays on your machine. API keys, conversations, memories, and skills live locally in ~/.evoforge. Nothing leaves except the calls to the model provider you configured.

Execution has boundaries. Three sandbox levels (read-only / workspace-write / full access), mandatory confirmation for destructive operations, hard refusal of dangerous command patterns, and an audit trail for everything it runs. Hand it a long task and go make coffee.

📦 Download

Grab the latest version from GitHub Releases:

PlatformDownload
Windows.exe installer
macOS.dmg image

After installing, Help → Check for Updates keeps you current.

🚀 Three steps in

  1. Pick a model — choose a provider and paste an API key on the launch page (or hit Set up later and look around first)
  2. Give it work — from "organize this folder" to "write tests for this repo and open a PR"
  3. Watch it evolve — after a few tasks, open the Skills panel and see what it has learned

💬 Feedback

Found a bug or want a feature? Open an Issue — attaching relevant output from the Logs panel makes diagnosis much faster.


EvoForge © 2026 · Non-commercial license · Evolution against forgetting

Frequently Asked Questions

What is EvoForge?

EvoForge is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by smallchao. 一个高性能的具备进化能力的桌面智能体工作台. It has 103 GitHub stars.

Is EvoForge safe to use?

Yes. EvoForge passed SkillsLLM's automated security scan — a dependency vulnerability audit plus prompt-injection heuristics — with no high-severity issues. You can read the full report in the Security Report section on this page.

How do I install EvoForge?

Clone the repository with "git clone https://github.com/smallchao/EvoForge" and add it to your Claude Code skills directory (see the Installation section above).

Are there alternatives to EvoForge?

Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh EvoForge against similar tools.

Comments (0)

No comments yet. Be the first to share your thoughts!

ECC

by affaan-m

10

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

242,21936,702JavaScript
AI Agentsai-agentsanthropicclaude-code
View details
15

An agentic skills framework & software development methodology that works.

234,96620,863Shell
AI Agentsai-agentsbrainstorming
View details

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

185,94028,768JavaScript
AI Agentsai-agentsanthropicclaude-code
View details

cc-switch

by farion1231

3

A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io

128,8688,826Rust
AI Agentsclaude-codeai-tools
View details

claude-code

by anthropics

Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.

120,03119,897Shell
AI Agents
View details

Developers Also Liked

Based on votes and bookmarks from developers who liked this skill

ECC

by affaan-m

10

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

242,21936,702JavaScript
AI Agentsai-agentsanthropicclaude-code
View details
15

An agentic skills framework & software development methodology that works.

234,96620,863Shell
AI Agentsai-agentsbrainstorming
View details

n8n

by n8n-io

12

Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.

201,88160,308TypeScript
MCP Serversapisai-tools
View details

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

185,94028,768JavaScript
AI Agentsai-agentsanthropicclaude-code
View details

cc-switch

by farion1231

3

A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io

128,8688,826Rust
AI Agentsclaude-codeai-tools
View details