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/developerz-ai/ui-debugger-mcp/claude-mdgit clone --depth 1 https://github.com/developerz-ai/ui-debugger-mcpWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/developerz-ai/ui-debugger-mcp/claude-md)<a href="https://agentmods.dev/instructions/developerz-ai/ui-debugger-mcp/claude-md"><img src="https://agentmods.dev/badge/instructions/developerz-ai/ui-debugger-mcp/claude-md.svg" alt="Measured on agentmods" height="20"></a>What 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 | $0.03365 | $0.03365 |
| Opus 5 | $0.01682 | $0.01682 |
| Sonnet 5 | $0.00673 | $0.00673 |
| Haiku 4.5 | $0.00336 | $0.00336 |
Grade B, and why
ui-debugger-mcp CLAUDE.md scanned grade B with 1 finding 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 3d ago.
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.
Reads MCP configurationmediumAgent snooping
mcp.json carries server URLs and auth tokens; reading it lets a mod discover and abuse other integrations.
- `.mcp.json` — how to LAUNCH server (command, model API key + base url). Gitignored. Secret. How it starts
The opening of the file, as written. The whole thing — 247 lines — stays where its author put it; the contents beside it link to each section on GitHub.
UI Debugger MCP
Bun · TypeScript · Vercel AI SDK · OpenAI-compatible router (OpenRouter default). MCP server that debugs UIs autonomously.
Smart agent hands a goal (a "story") to a small fast agent inside this server. The small agent drives the target, gathers evidence, reports findings. Smart agent fixes code, asks again. Loop until the UI works. No human clicking.
Goal: stuff WORKS + LOOKS NICE. Three actors (docs/idea/models.md):
- smart agent — the boss (Claude/caller): sets goals, fixes code, loops.
- fast guy — the driver (fast, text-only, blind): controls the target.
- vision guy — the eyes (multimodal): describes screenshots, judges looks.
Response Rules
- Execute. No preamble.
- Lead with action or answer.
- Terse. Fragments OK.
What this is
- MCP server (stdio). Ships as npm package, run via
npx/bunx. - The agent brain runs inside the server (Vercel AI SDK). Not the caller's.
- Session keyed by current dir (cwd). One project = one session.
- One debug run at a time (persistent Chrome profile locks the profile).
Reference repos (read for patterns, don't copy blind)
../ai-task-master— build template: Bun + TS + Vercel AI SDK, subagent-as-tool, MCP, Zod, Biome.../gold-standards-in-ai/docs— house rules: few fat tools, SRP, custom errors, compressed config.../../sebyx07/claude-code-bible/docs— deeper handbook.?debug-ai=truelogin bypass pattern.
Architecture
src/main.ts— boot stdio MCP server.src/mcp/— MCP server, tool defs (few fat tools, NOT one-per-action).src/agent/— debug agent (Vercel AI SDK loop). The fast guy (driver) + the vision guy (eyes vialook). Models per role via OpenAI-compatible router (OpenRouter default). Defaults: deepseek (text), glm (image).prompts/— OUR system prompts: in-repo, versioned, tested, provider-agnostic. NEVER rely on a 3rd-party model's defaults. Written compressed — they are resent every step (../../sebyx07/claude-code-bible/docs/11-compressed-config.md).
src/adapters/— target control behind one shared contract.browser/— web via CDP (headless default). ✅ shipped.desktop/— X11/Wayland windows. Covers desktop app AND mobile emulator. ✅ shipped.android/— ADB + uiautomator. ✅ shipped.
src/session/— cwd-keyed session, per-project workspace, state.src/config/— load.ui-debugger-mcp.json, resolve model/targets.src/services/— business logic. Thin handlers, logic here.
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.
- 3d ago First seen · 247 lines · 3,365 tokens per session scan B a765d5d23b7a
ui-debugger-mcp CLAUDE.md is an instructions file published in the GitHub repository developerz-ai/ui-debugger-mcp (2 stars, last pushed 27d ago), licensed MIT. It adds 3,365 tokens to every session, about $0.0168 per session on Opus 5. A static security scan graded it B with 1 finding (reads mcp configuration). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
bxc CLAUDE.md
Instructions for aphrody-code/bxc, covering claude.md — bxc, objectif directeur (depuis 2026-08-27), rappels critiques, commandes essentielles and commandes des scrapers dédiés.
bxc GEMINI.md
Instructions for aphrody-code/bxc: This repository's operating guide is CLAUDE.md — read it first.
obscura AGENTS.md
AGENTS.md instructions for h4ckf0r0day/obscura, covering agents.md, build, rendering and stealth, no rendering, with rustls or stealth and test.
browser-harness AGENTS.md
AGENTS.md instructions for browser-use/browser-harness, covering code priorities, commands, doctor — install/daemon/browser state, unit tests (no live browser) and security.
browser-harness CLAUDE.md
Claude Code instructions for browser-use/browser-harness, a project described as: Browser Harness | Self-healing harness that enables LLMs to complete any task.
terminal AGENTS.md
Instructions for browser-use/terminal, covering agent notes, rust rewrite verification loop and terminal ui testing standard.