Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/williamcorrea23/androidbuilder-mcp/agents-mdgit clone --depth 1 https://github.com/williamcorrea23/androidbuilder-mcpWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00523 | $0.00523 |
| Opus 5 | $0.00262 | $0.00262 |
| Sonnet 5 | $0.00105 | $0.00105 |
| Haiku 4.5 | $0.00052 | $0.00052 |
Grade A, and why
androidbuilder-mcp AGENTS.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
Copies of this mod
1 near-identical copy found in the catalogue:
- androidbuilder-mcp CLAUDE.md — 98% identical, 2 lines differ
What it actually says
AGENTS.md — AndroidBuilder MCP Server
Project Overview
This is an MCP (Model Context Protocol) server written in TypeScript/Node.js that bridges LLMs to the MIT App Inventor / AndroidBuilder / Niotron / Kodular web IDEs via Chrome's DevTools Protocol (CDP) using Puppeteer Core.
Architecture
src/
├── index.ts # MCP Server, tool registration, routing
├── tools.ts # All tool implementations (10 tools)
└── browser-bridge.ts # CDP connection, page cache, timeout handling
Key Design Decisions
- Single cached CDP connection — avoids WebSocket leak where each call opened a new connection
- 3-strategy property update — GWT API → top-frame API → DOM input field
- Timeout on
evaluateInPage— prevents hanging if IDE page freezes (default 15s) - Graceful SIGTERM/SIGINT disconnect — via
bridge.setupProcessHandlers() - Env-var driven config —
BROWSER_DEBUG_PORT,EVAL_TIMEOUT_MS,ANDROIDBUILDER_DEBUG
Tools (10 total)
| Tool | View Required |
|---|---|
get_project_info |
Any |
get_component_tree |
Designer |
get_component_properties |
Designer (component selected) |
update_component_property |
Designer |
add_blocks |
Blocks |
get_blocks_xml |
Blocks |
clear_blocks |
Blocks |
switch_screen |
Any |
list_screens |
Any |
click_element |
Any (fallback) |
Development
npm run build # compile TypeScript → dist/
npm start # run compiled server
ANDROIDBUILDER_DEBUG=1 npm start # verbose logging
Common Errors & Fixes
ECONNREFUSED→ Chrome not started with--remote-debugging-port=9222No active tab found→ IDE not open or project not loadedgwt-Tree not found→ Switch to Designer viewBlockly not available→ Switch to Blocks view- Timeouts → Increase
EVAL_TIMEOUT_MS, reload IDE page
Supported IDEs
androidbuilder.in, ai2.appinventor.mit.edu, code.appinventor.mit.edu, niotron.com, makeroid.io, appybuilder.com
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 51 lines · 523 tokens per session scan A 6800f7b9e5a1
androidbuilder-mcp AGENTS.md is an instructions file published in the GitHub repository williamcorrea23/androidbuilder-mcp (0 stars, last pushed 26d ago), licensed MIT. It adds 523 tokens to every session, about $0.0026 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
rustwright AGENTS.md
Instructions for Skyvern-AI/rustwright, covering repository instructions, public boundary, pull-request review, local verification and browser launches.
browser-debugger-cli CLAUDE.md
Instructions for szymdzum/browser-debugger-cli, covering claude.md, agent-friendly discovery (start here), essential patterns, commandrunner (src/commands/shared/commandrunner.ts) and error handling.
cdpilot CLAUDE.md
Instructions for mehmetnadir/cdpilot, covering cdpilot, kimlik, hızlı başlangıç, dosya haritası and dikkat edilecekler.
cdpilot GEMINI.md
Instructions for mehmetnadir/cdpilot, covering context-mode — mandatory routing rules, blocked commands — do not attempt these, curl / wget — blocked, inline http — blocked and webfetch / web browsing — blocked.
browsertap-mcp AGENTS.md
Instructions for LinVireo/browsertap-mcp, covering agents.md, 3. tab ids are not stable -- never remember one, 4. screen-coordinate tools act on the visible tab, 5. mv3 chrome.alarms has a 60-second floor and 6. the http port is token-authenticated.
chatgpt-web-handoff CLAUDE.md
Instructions for sologuy/chatgpt-web-handoff, covering claude.md —— claude code 专用补充, 浏览器通道, 上下文纪律, 这个 chrome 是用户日常在用的 and skill 真相源在仓库.