SuperAPI AI Directory
Partner with us中文AI newsFavorites

AI coding

2,798 results · Page 76 of 78

Ptah

AI coding

Ptah is an AI coding extension for Visual Studio Code, centered on an "AI coding symphony," namely vendor-agnostic AI orchestration combined with intelligent workspace analysis and multi-agent collaboration. It can scan and understand a user's codebase, automatically configure specialized AI agents, and schedule them to handle complex development tasks in parallel. The extension includes a built-in MCP server that provides 14 API namespaces, giving agents structured access to workspace analysis, LSP diagnostics, AST parsing, file search, dependency graphs, and editor state. It supports automatically identifying built-in agents from installed CLIs, configuring custom agents with your own providers and API keys, and generating, monitoring, guiding, and collecting background agent results through MCP tools. Ptah provides seven categories of orchestration workflows (feature development, bug fixing, refactoring, documentation, research, DevOps, and creative), with user approval checkpoints at each stage, along with native editor chat, execution tree visualization, session management, cost tracking, and a performance dashboard. It is licensed under FSL-1.1-MIT, which automatically converts to a full MIT license two years after release.

Read more

Quack AI

AI coding

Quack AI is an open-source tool that turns team contribution guidelines into contextual assistance within the editor. By parsing existing documentation and past PR comments, it helps maintainers organize expectations around naming, error handling, testing, review etiquette, and more into structured guidelines. When developers write code in the IDE, the extension extracts only the rules relevant to the currently open file, checks code diffs, and generates code that conforms to the guidelines, so issues can be caught before review. The tool runs on local LLMs and can be self-hosted via Ollama and any open-source model, keeping source code on your own infrastructure. Maintainers can also use the web app to see which rules are violated most and which rules no one understands, thereby optimizing the contribution process. Note: the product has been sunsetted, but the code is open source on GitHub under the Apache 2.0 license and can be freely forked, run, and reused. The project consists of three parts: the API (guideline storage, retrieval, and LLM orchestration), the editor extension, and the web app. The founding team has left the developer tools space and is now building Rimward (solar site safety monitoring).

Read more

Retained mode documentation

AI coding

Roo Code's Modes are specialized personas used to let the AI assistant adjust its behavior according to the current task. Each mode provides different capabilities, expertise, and access levels to help accomplish specific goals. Modes remember the last used model, automatically select it when switching, and persist across sessions. Built-in modes include Code (default), Ask, Architect, Debug, and Orchestrator (also known as Boomerang). Users can switch modes via the dropdown menu, slash commands (such as /architect), keyboard shortcuts (macOS: ⌘ + .; Windows/Linux: Ctrl + .), or by accepting suggestions. Modes support customization, allowing you to define tool access, file permissions, and behavioral instructions. Tool groups include read (file reading, listing, searching), edit (file modification and creation), command (terminal command execution), and mcp (Model Context Protocol server interaction).

Read more

ReviewPhin

AI coding

ReviewPhin is a self-hosted code review automation tool for GitLab merge requests and GitHub pull requests. It is deployed as a container, runs via Docker Compose or a published Helm Chart, and the same image provides the review worker, the initial setup process, and documentation. Users can register GitLab or GitHub platform credentials and onboard projects or repositories as tenants; on the model side, they can use the built-in Copilot CLI path, or create named model configurations for OpenAI-compatible, Azure, Anthropic, and other endpoints. Storage can start with SQLite, switch to Flotiq when an admin panel is needed, or load a custom storage adapter. After a review is triggered, ReviewPhin uniformly coordinates review findings, replies, summaries, and status, helping engineers focus review time on design, risk, and intent rather than hunting for typos and obvious issues. In production, /app/data and /app/tmp must be persisted.

Read more

Reviu

AI coding

Reviu is a native desktop application designed for reviewing code written by AI coding agents. It can launch command-line agents from the ACP registry that you have already installed and logged into, such as Claude Code, Codex, and Gemini, letting you watch the agent's running status in real time in a local session, review the code diffs it produces line by line, and merge your revision feedback into a single prompt sent back to the agent via inline comments. Once confirmed correct, you can perform real Git operations such as staging, rebasing, committing, and pushing directly within Reviu. Reviu is built with Rust and GPUI, does not depend on Electron or a webview, runs agents as local processes, and keeps code and data on your machine by default. Local agent sessions, local diff review, inline comments, parallel worktrees, and local Git workflows are all free; the paid Reviu Pro tier adds GitHub integration, including a pull request dock, GitHub review comments, checks, merge operations, and a notification inbox. Pro offers a 14-day free trial. Reviu does not meter agent usage or require pasting API keys, and works with your existing CLI subscription.

Read more

Robomonkey

AI coding

Robomonkey is an AI-powered Chrome extension building tool that lets users chat with AI in a sidebar and describe their needs in natural language to generate and install browser extensions, automation scripts, and data scraping tools without writing code. The product supports desktop only, offers a free trial and a limited-time one-month free offer, and has a credit-based paid plan. According to the official page, its Chrome Web Store rating is 4.8 stars (40 reviews).

Read more

SeaGOAT

AI coding

SeaGOAT is a local code search engine for the AI era, using vector embeddings to enable semantic search of codebases. It runs the SeaGOAT server entirely on the local machine, does not rely on third-party remote APIs, and uses ChromaDB, a local vector database, together with ripgrep, balancing semantic matching with regex/keyword search. Installation requires Python 3.10 or newer, and optionally and recommendedly installing bat for colorized display of results. After starting the server, you can query repositories with the gt or seagoat commands, and enhance results through YAML configuration, .seagoat.yml project configuration, and an optional --generative flag (local Ollama by default, and also configurable for a cloud LLM). It supports a variety of text and programming language files, ignores .gitignore and binary files, is suitable for quickly finding code in local private repositories, and is designed not to block users' daily use.

Read more