6 Must‑Try Open‑Source AI & Automation Tools to Supercharge Your Workflow

AutoFigure‑Edit – Automatic Academic Figure Generation

AutoFigure‑Edit, developed by a team at Westlake University and selected for ICLR 2026, converts textual descriptions of methods sections directly into editable SVG diagrams, eliminating manual layout work.

Core workflow (four steps): 1) Use a large language model (LLM) to generate a sketch from text; 2) Apply SAM‑3 for image segmentation to identify icon regions; 3) Employ RMBG‑2.0 for high‑precision background removal (90.14% accuracy); 4) Automatically assemble the final vector graphic.

The tool also supports style transfer—upload a reference image and the generated figure adopts that style, ensuring visual consistency across a paper.

Open‑source repository: https://github.com/ResearAI/AutoFigure-Edit

WebRPA – Zero‑Code Web Automation

WebRPA offers a visual, drag‑and‑drop workflow builder that lets non‑programmers automate web tasks without writing Python code.

Key features include 260 built‑in modules covering web interaction, data extraction, file handling, media conversion, and AI dialogue. The package bundles Python 3.13 and Node.js, so it runs out‑of‑the‑box after extraction.

Browser automation is powered by Playwright, supporting CSS selectors, XPath, clicks, inputs, screenshots, etc. Data handling includes JSON parsing, regular expressions, Excel read/write, with export options to files or MySQL. Media processing relies on FFmpeg for over 50 audio/video formats. AI integration works with OpenAI, Zhipu, Tongyi Qianwen and other providers.

A standout feature is the trigger system, offering 10 trigger types such as Webhook, scheduled tasks, file monitoring, and hotkey listeners, enabling complex scenarios like price monitoring, news scraping, or auto‑login.

Open‑source repository: https://github.com/pmh1314520/WebRPA

Trellis – AI Code‑Generation Efficiency Toolkit

Trellis addresses the repetitive task of re‑describing project conventions to AI code assistants (Claude Code, Cursor, Codex, etc.). It stores specifications (file naming, component standards, comment guidelines) in a Markdown spec directory and automatically injects them at the start of each AI conversation.

Additional capabilities include parallel sessions across multiple worktrees, a /trellis:record-session command to log work sessions, and a layered architecture that loads only the relevant specifications for the current task, reducing token usage and improving generation accuracy.

Open‑source repository: https://github.com/mindfold-ai/Trellis

qqbot – Quick QQ Bot Integration via OpenClaw

The qqbot plugin for the OpenClaw framework enables developers to create personal QQ chatbots with minimal configuration.

Setup steps: 1) Register an account on the QQ Open Platform and create a bot application to obtain AppID and AppSecret; 2) Install the plugin with the command openclaw plugins install @sliverp/qqbot@latest and provide the credentials.

The bot currently supports private‑chat interactions, offering features like intelligent Q&A, task reminders, and simple command execution.

Open‑source repository: https://github.com/sliverp/qqbot

CountBot – Production‑Grade AI Agent Framework

CountBot, released in February 2026, is a 21 K‑line Flutter project that implements a “tool‑as‑agent” architecture. Although marketed as a word‑count utility, its modular design allows easy extension to AI text analysis, writing assistance, and content moderation.

Key design principles: each capability is encapsulated as an independent agent module, enabling clean separation of concerns and industrial‑grade code quality. The framework runs on Windows, macOS, Linux, and HarmonyOS 6.0.

For developers interested in AI agent architecture and cross‑platform Flutter development, CountBot serves as a concrete reference.

Open‑source repository: https://github.com/countbot-ai/CountBot

ai‑openclaw‑skeletons – Ready‑Made AI Agent Development Templates

Part of the OpenClaw ecosystem, the ai-openclaw-skeletons repository provides plug‑and‑play AI Agent skeletons that model real‑world employee collaboration: Packs (skill units), Bundles (role compositions), Hooks (supervision), and Contracts (coordination rules).

The system is ecosystem‑agnostic, built on long‑lived engineering patterns such as Skill, MCP, CLI, Hook, and Cron, allowing rapid assembly of production‑grade AI agents without starting from scratch.

Open‑source repository: https://github.com/1596941391qq/ai-openclaw-skeletons

Practical Tips for Using These Projects

Always read the official README to understand the latest features and configuration requirements.

For environment‑heavy tools (WebRPA, qqbot), prefer the provided packaged versions to avoid local dependency conflicts.

Leverage the visual editors in AutoFigure‑Edit and WebRPA to focus on outcomes rather than manual code edits.

These open‑source tools are free and actively maintained, offering researchers, office workers, and developers a way to automate repetitive tasks and accelerate AI‑driven development.