🔨 A2A-Forge

The desktop workbench for testing and debugging A2A protocol agents

中文 | English

---

_{Three-panel layout: agents sidebar · skill browser · test panel with live results}

_{Dark mode with system-aware theme detection}

What is A2A-Forge?

A2A-Forge is a native desktop application for developers building A2A (Agent-to-Agent) protocol agents. Think of it as Postman for A2A — add any agent by URL, browse its skills, and run live test interactions without writing code.

Built with Tauri 2 for a fast, lightweight, cross-platform experience.

Features

• 🔍 Agent Discovery — Fetch and inspect .well-known/agent.json agent cards instantly
• ⚡ Skill Testing — Invoke any skill with adaptive input (text, JSON, file upload)
• 📡 Live Streaming — SSE-based streaming with real-time status updates for async tasks
• 🔐 Secure Auth — Per-agent default headers with OS keychain credential storage
• 🖼️ Smart Preview — Auto-detect and render images, video, and audio inline
• 📋 Curl Export — Copy equivalent curl commands with one click
• 📜 History — Searchable execution history with saved test cases
• 🗂️ Workspaces — Organize agents into separate workspaces
• 🌗 Themes — System-aware dark/light mode with manual override
• ⌨️ Keyboard First — Ctrl+N add agent · Ctrl+Enter run · Ctrl+Shift+C copy curl

Quick Start

Prerequisites

• Rust (stable toolchain)
• Node.js v18+
• Platform dependencies: see Tauri Prerequisites

Install & Run

git clone https://github.com/learningpro/a2a-forge.git
cd a2a-forge
npm install
npm run tauri dev

Build for Production

npm run tauri build
# Output: .dmg (macOS) / .msi + .exe (Windows)

Usage

1. Add an Agent

Click + Add agent card or press Ctrl/Cmd+N. Enter the agent's base URL — A2A-Forge fetches the card from /.well-known/agent.json and displays all available skills.

2. Configure Auth

Click the ⚙️ gear icon next to the agent name to set default headers (e.g., X-API-Key). These persist across sessions and apply to all skill tests for that agent.

3. Test a Skill

Select a skill → write your input → click Run (or Ctrl/Cmd+Enter). The response viewer renders results with smart media detection — images display inline, videos play in-app, JSON is syntax-highlighted.

4. Review & Iterate

Every execution is saved to history. Save frequently-used requests as named test cases for one-click re-run. Copy the equivalent curl command to share with teammates.

Tech Stack

Roadmap

v0.1 — Current Release ✅

Agent card management, skill browser, adaptive test input, async task execution with auto-polling, SSE streaming, response viewer with smart media preview, test history, saved test cases, per-agent auth headers, workspaces, settings, keyboard shortcuts.

v0.2 — Automated Testing ✅

• Test suites — group test cases into automated sequences (agent-level + workspace-level)
• Assertions — 7 types: status_equals, json_path_equals/exists/contains/matches, response_time_lt, contains_media
• CI integration — run test suites from command line
• Test reports — export results as HTML/JSON

v0.3 — Local Registry Proxy ✅

• Local A2A registry — embedded axum proxy for agent discovery
• Request/response interception and modification
• Latency simulation and fault injection
• Traffic recording and replay

v0.4 — Community Hub ✅

• Community agent directory — discover and load popular A2A agents
• Shared test collections — import/export community test suites
• Agent health monitoring — periodic card refresh with alerts
• Favorites — star and organize frequently-used agents

v0.5 — Advanced Workspace ✅

• Workspace sharing — export/import full workspace configs
• Environment variables — per-workspace variable substitution with secret masking
• Request chaining — pipe output of one skill into another with JSONPath extraction
• Diff view — compare responses across runs

Contributing

Contributions are welcome! Please open an issue or submit a pull request.

npm run tauri dev        # Start dev server
npx tsc --noEmit         # Type check
npx vitest run           # Run tests

License

MIT — Copyright 2026 Orange Dong