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.
git clone --depth 1 https://github.com/ncklrs/claude-chrome-user-testingWrote 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/commands/ncklrs/claude-chrome-user-testing/annotate)<a href="https://agentmods.dev/commands/ncklrs/claude-chrome-user-testing/annotate"><img src="https://agentmods.dev/badge/commands/ncklrs/claude-chrome-user-testing/annotate/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/ncklrs/claude-chrome-user-testing/annotate"><img src="https://agentmods.dev/badge/commands/ncklrs/claude-chrome-user-testing/annotate.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00031 | $0.01301 |
| Opus 5.5 | $0.00012 | $0.00520 |
| Sonnet 5.5 | $0.00006 | $0.00260 |
| Haiku 4.5 | $0.00003 | $0.00130 |
Grade A, and why
annotate 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 1mo 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.
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.
How it starts
The opening of the file, as written. The whole thing — 199 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Annotate Command
Add visual annotations to the current browser page and capture an annotated screenshot.
Arguments
Parse the following from $ARGUMENTS:
--element <selector>(required): CSS selector or description of the element to annotate--type <type>(optional): Annotation type -box,highlight, orcallout(default: box)--color <color>(optional): Color -red,orange,green,blue(default: red)--style <style>(optional): Border style -solidordashed(default: solid)--label <text>(optional): Label text to display above the annotation--number <n>(optional): Callout number (for callout type)--filename <name>(optional): Screenshot filename (default: annotated-{timestamp}.png)
Prerequisites
This command requires:
- Claude running with
--chromeflag - A page already loaded in the browser
If no page is loaded, inform the user:
No page is currently loaded. Navigate to a URL first with:
browser_navigate to https://example.com
Annotation Flow
Step 1: Find Target Element
Use browser_snapshot to get the current page accessibility tree. Find the element matching the user's description or selector.
If using a CSS selector:
- Try to match directly
- If not found, suggest alternatives
If using a description (e.g., "submit button"):
- Search the snapshot for matching elements
- Look for buttons, links, or inputs with matching text/labels
Step 2: Get Element Position
Use browser_evaluate to get the element's bounding box:
Get the bounding rectangle of the element matching: [selector]
Return: { x, y, width, height } in page coordinates
Step 3: Create Annotation Overlay
Use browser_evaluate to create the annotation container:
- Create a div with id "uta-overlay"
- Position it absolutely over the page
- Set pointer-events to none
- Set z-index to 99999
Step 4: Add Annotation
Based on the --type argument:
Box (default):
- Create a div with colored border
- Position at element coordinates
- Add label if provided
- Use dashed border if
--style dashed
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.
- 1mo ago First seen · 199 lines · 31 tokens per session scan A 6546c7806870
annotate is a command published in the GitHub repository ncklrs/claude-chrome-user-testing (21 stars, last pushed 9mo ago), licensed MIT. It adds 31 tokens to every session and 1,301 once invoked, about $0.0001 per session on Opus 5.5. 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 2026-08-30.
Other commands, from other repositories
anti-slop-designer
Use when a UI looks machine-made rather than decided. Violet gradients, glassmorphism everywhere, identical cards in a grid, untouched shadcn or Material defaults, emoji bullets, or copy stuffed with seamless and unlock.
svg-create
Author or clean up an SVG asset — icon, illustration, mascot pose, logo mark — so it scales from viewBox, recolors from tokens, animates without a rewrite, is optimized, and has an accessible name.
agentation-fix
Session-2 fix loop for Agentation — read structured annotations from the dev overlay and apply targeted UI fixes.
scan-ai-tells
Scan UI or marketing copy for AI-default tells and content-authenticity misses — deletion list, not a redesign brief.
design-critic
Use when you want an honest read on a design instead of encouragement. What is wrong, ranked by severity, with the reason attached. Works on a screenshot, a mockup, or a described screen.
interaction-designer
Use when deciding how something should behave rather than how it looks. Click and tap targets, gestures, keyboard paths, loading and empty and error states, and the feedback each action gives.