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/tan-yong-sheng/ai-vision-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/commands/tan-yong-sheng/ai-vision-mcp/audit-accessibility)<a href="https://agentmods.dev/commands/tan-yong-sheng/ai-vision-mcp/audit-accessibility"><img src="https://agentmods.dev/badge/commands/tan-yong-sheng/ai-vision-mcp/audit-accessibility/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/tan-yong-sheng/ai-vision-mcp/audit-accessibility"><img src="https://agentmods.dev/badge/commands/tan-yong-sheng/ai-vision-mcp/audit-accessibility.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 | $0.00025 | $0.00759 |
| Opus 5 | $0.00013 | $0.00380 |
| Sonnet 5 | $0.00005 | $0.00152 |
| Haiku 4.5 | $0.00003 | $0.00076 |
Grade A, and why
audit-accessibility 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 11d 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 — 68 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/design-eval:audit-accessibility
Deep accessibility compliance review with WCAG 2.1 or WCAG 3.0 assessment. Supports two modes: automated axe-core scanning or combined axe-core + AI analysis.
Arguments
| Argument | Description | Example |
|---|---|---|
--imageSource |
URL, file path, or base64 image to evaluate | https://example.com/form.jpg |
--mode |
Scan mode: quick (axe-core only) or deep (axe-core + AI analysis, default) |
--mode quick |
--level |
WCAG compliance level: A (minimum), AA (standard), AAA (enhanced) | --level AA |
--wcag-version |
WCAG standard: 2.1 (compliance-focused) or 3.0 (outcome-focused) | --wcag-version 3.0 |
--email |
Email for login (optional, if page requires authentication) | --email [email protected] |
--password |
Password for login (optional, if page requires authentication) | --password MyPassword123 |
--design-system |
Path to DESIGN.md file for design-aware remediation (optional) | --design-system ./DESIGN.md |
--userPrompt |
Additional focus areas or custom instructions | --userPrompt "check keyboard navigation for power users" |
Mode Options
--mode quick — Fast automated scanning
- Runs axe-core only (no API calls)
- Immediate results
- Catches low-hanging fruit (contrast, labels, semantic HTML)
- Output: Markdown report + JSON findings
- Best for: Quick feedback, CI/CD pipelines, catching obvious issues
--mode deep (default) — Comprehensive analysis
- Runs axe-core first (automated findings)
- Sends findings + screenshots to AI for deeper analysis
- AI provides context, remediation code, design patterns
- Output: Combined report (automated + AI insights)
- Best for: Full audits, design-aware fixes, actionable guidance
Examples
# Quick axe-core scan only (no AI)
/design-eval:audit-accessibility --imageSource https://example.com --mode quick
# Deep analysis (axe-core + AI, default)
/design-eval:audit-accessibility --imageSource https://example.com --mode deep --level AA
# With authentication
/design-eval:audit-accessibility --imageSource https://example.com/dashboard --mode deep --email [email protected] --password &Test1234
# WCAG 3.0 with design system reference
/design-eval:audit-accessibility --imageSource ./form.png --wcag-version 3.0 --design-system ./DESIGN.md
# Quick scan with custom focus
/design-eval:audit-accessibility --imageSource https://example.com --mode quick --userPrompt "check keyboard navigation and skip links"
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.
- 11d ago First seen · 68 lines · 25 tokens per session scan A d7fa583ec31c
audit-accessibility is a command published in the GitHub repository tan-yong-sheng/ai-vision-mcp (78 stars, last pushed 5mo ago), licensed MIT. It adds 25 tokens to every session and 759 once invoked, about $0.0001 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 2026-08-30.
Other commands, from other repositories
fidelity
Measure how faithfully a clone reproduces a site — pixel-diff plus motion-fidelity into one 0-100 score, a letter grade, a ranked correction plan, and a shareable card.
qa
Verify the current work against the hard rules and the design.
bakeoff
Calibration loop for taste-heavy choices (v7.16.0). Generates N candidate implementations/outputs, renders a side-by-side contact sheet, captures your verdict, pins the winner with a regression test, and distills the ruling into the taste model. Converts fuzzy judgment into deterministic constants.
stark-dogfood
Use this command when testing whether Stark actually generates better designs, refreshing proof screenshots, creating showcase examples, or turning generated-output failures into guidance changes.
review
Review the current git diff — auditor selected by --mode (general, security, a11y, deps, perf, tests, contributor-trust, ux) or auto-inferred from the diff when omitted. Pass a PR number to review a specific PR; use --check-followup to re-check a PR after the owner has addressed feedback.
p3-ux-a11y
Defines WCAG target, contrast requirements, keyboard navigation, screen reader requirements and testing methods.