Skills accessibility-engineering
๐Ÿ“ฆ

accessibility-engineering

Content revision r1 Medium Risk โš™๏ธ External commands๐Ÿ“ Filesystem access๐ŸŒ Network access

Audit and Fix Web Accessibility Issues

Automated checks miss barriers that affect users navigating by keyboard or screen reader. Combine scanner findings with manual tests to prioritize fixes and verify them.

Supports: Claude Codex Code(CC)
Maintain this Skill? Claim it โš ๏ธ 50 Poor

Install with my Agent

Copy this request to your Agent. It includes the canonical Skill page and manifest.

Agent request
Review the Skillstore skill "accessibility-engineering" from https://skillstore.io/skills/addyosmani-accessibility-engineering.md and its manifest at https://skillstore.io/api/skills/addyosmani-accessibility-engineering/manifest. Verify the artifact. You may proceed after verification, subject to the environment's own policy.

Your Agent should still show its plan and request any confirmation required by the security policy.

Test it

Using "accessibility-engineering". Review a form with an unlabeled email input and an error message shown only in red.

Expected outcome:

  • The email input needs an associated visible label so assistive technology can announce its purpose.
  • Add error text and associate it with the input. Verify the error without relying on color.
  • Screen reader behavior remains untested until someone checks it with a named browser and screen reader.

Using "accessibility-engineering". Summarize an audit in which axe completed but pa11y did not return a valid report.

Expected outcome:

  • The URL audit is incomplete. Axe completed, but pa11y did not produce a valid report.
  • Do not report a combined violation count. Resolve the pa11y failure and rerun the audit.

Security Audit

Medium Risk
v1 โ€ข 11/19/2026 Open versioned report

Most alerts describe Markdown formatting, documentation links, or routine report handling. The script embeds environment-derived paths in JavaScript, automatically runs npm packages, and opens user-supplied targets with no enforced audit scope. Page content enters the reports, but the reviewed files show no evidence of deliberate data exfiltration or embedded prompt injection.

2
Files scanned
368
Lines analyzed
15
Review items
0
False positives ignored

Confirmed security concerns (3)

Medium
Automatic Execution of Downloaded Audit Tools
The workflow can download and run npm tools with no installation prompt. Top-level version pins do not replace a reviewed dependency lockfile.
The examples and wrapper use npx with automatic confirmation. The reviewed bundle contains no lockfile or preapproved local tool setup.
Medium
Browser Audit Does Not Enforce Target Scope
The wrapper passes user-supplied URLs and paths to browser tools without enforcing an approved scope. Scans can reach resources available to the runner.
The wrapper checks for missing input and then forwards targets unchanged. The resources reachable in practice depend on the runner and browser configuration.
Low
Page-Controlled Content Enters Audit Reports
Audit reports can include sensitive markup or page text that looks like instructions. Assistants and report recipients should treat those excerpts as untrusted data.
The JSON output preserves axe node HTML and pa11y context without filtering. No evidence found of an embedded attack or automatic report upload.
Capability review items (15)

These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.

Medium
Node.js fs operations
const result = JSON.parse(fs.readFileSync('$TEMP_DIR/result.json', 'utf8'));
The result path is inserted into JavaScript passed to Node. An unsafe temporary directory path can change the code instead of supplying a filename.
Medium
Node.js fs operations
const axe = JSON.parse(fs.readFileSync('$TEMP_DIR/axe-report.json', 'utf8'));
The axe report path is expanded inside executable JavaScript. A path inherited through TMPDIR can break the intended string literal.
Medium
Node.js fs operations
fs.writeFileSync('$TEMP_DIR/result.json', JSON.stringify(result, null, 2));
The write uses a path embedded in Node source rather than an argument. Unsafe temporary path characters can alter that source.
Medium
Node.js fs operations
const result = JSON.parse(fs.readFileSync('$TEMP_DIR/result.json', 'utf8'));
The pa11y merge inserts the result path into Node source. That path can be influenced through TMPDIR and is not escaped for JavaScript.
Medium
Node.js fs operations
const pa11y = JSON.parse(fs.readFileSync('$TEMP_DIR/pa11y-report.json', 'utf8'));
The pa11y report path is expanded into a JavaScript string literal. Parsing the report as JSON does not protect the surrounding source.
Medium
Node.js fs operations
fs.writeFileSync('$TEMP_DIR/result.json', JSON.stringify(result, null, 2));
The merged report write places the temporary path directly in Node source. Serializing the result as JSON does not escape the path.
Medium
Node.js fs stat operations
if (fs.statSync('$TEMP_DIR/axe-report.json').size > 0) {
The axe fallback size check embeds the temporary path in Node source. Node parses that source even if the fallback never runs.
Medium
Node.js fs stat operations
if (fs.statSync('$TEMP_DIR/pa11y-report.json').size > 0) {
The pa11y fallback embeds the report path in the generated JavaScript. The path can alter that source before the fallback condition is checked.
Medium
Synchronous file operations
const result = JSON.parse(fs.readFileSync('$TEMP_DIR/result.json', 'utf8'));
The risk is path interpolation, not synchronous reading. The temporary path becomes part of the code that Node executes.
Medium
Synchronous file operations
const axe = JSON.parse(fs.readFileSync('$TEMP_DIR/axe-report.json', 'utf8'));
The axe input filename is expanded within the JavaScript program. The shell quoting does not escape it for a JavaScript literal.
Medium
Synchronous file operations
fs.writeFileSync('$TEMP_DIR/result.json', JSON.stringify(result, null, 2));
The synchronously written report is legitimate. The risk is that the destination path is interpolated into executable source.
Medium
Synchronous file operations
const result = JSON.parse(fs.readFileSync('$TEMP_DIR/result.json', 'utf8'));
The pa11y merge reads its prior result through a path embedded in Node source. The path inherits the generated temporary directory without JavaScript escaping.
Medium
Synchronous file operations
const pa11y = JSON.parse(fs.readFileSync('$TEMP_DIR/pa11y-report.json', 'utf8'));
The temporary report filename enters executable JavaScript rather than a separate argument. An unsafe path can change the parsed program.
Medium
Synchronous file operations
fs.writeFileSync('$TEMP_DIR/result.json', JSON.stringify(result, null, 2));
The pa11y merge write repeats the unescaped temporary path interpolation. The synchronous API itself is not the risk.
Medium
Synchronous file operations
const r = JSON.parse(require('fs').readFileSync('$RESULT_FILE', 'utf8'));
The text formatter inserts RESULT_FILE into Node source. The value derives from the temporary directory and is not escaped for JavaScript.
Audited by: codex
Share & cite this report

Share the versioned assessment report, neutral badge, embed card, and citations. Skillstore reports evidence without deciding whether this Skill is safe.

Open versioned report
Security Assessment

Copy report link

https://skillstore.io/skills/addyosmani-accessibility-engineering/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown badge

[![Skillstore security assessment](https://skillstore.io/badges/skills/addyosmani-accessibility-engineering/security.svg)](https://skillstore.io/skills/addyosmani-accessibility-engineering?utm_source=security_passport_badge)

HTML badge

<a href="https://skillstore.io/skills/addyosmani-accessibility-engineering?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/addyosmani-accessibility-engineering/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

Embed card

<iframe src="https://skillstore.io/embed/skills/addyosmani-accessibility-engineering.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>
Academic citations (APA ยท BibTeX ยท CFF)

APA citation

addyosmani. (2026). accessibility-engineering security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/addyosmani-accessibility-engineering/audits/1

BibTeX citation

@techreport{addyosmani-addyosmani-accessibility-engineering-2026, author = {addyosmani}, title = {accessibility-engineering security audit report (audit version 1)}, institution = {Skillstore}, year = {2026}, number = {1}, url = {https://skillstore.io/skills/addyosmani-accessibility-engineering/audits/1}, note = {Author version unspecified} }

CITATION.cff

cff-version: 1.2.0 message: "If you use this Skill, cite its author and this versioned security audit report." title: "accessibility-engineering security audit report (audit version 1)" version: "unspecified" type: report authors: - name: "addyosmani" date-released: "2026-11-19" url: "https://skillstore.io/skills/addyosmani-accessibility-engineering/audits/1" identifiers: - type: other value: "skillstore:addyosmani-accessibility-engineering:audit:1" description: "Skillstore immutable audit report identifier"

Skillstore Score

Why this score Evidence Confidence: Medium
45
Architecture
65
Maintainability
87
Content
65
Community
74
Spec Compliance

What You Can Build

Fix an inaccessible form

Verify missing labels and keyboard problems in an existing form. Make targeted fixes and retest the affected controls.

Verify a release candidate

Pair scanner reports with manual checks on a staging page. Record the browser, assistive technology, and checks that could not be run.

Prioritize accessibility findings

Review verified barriers and separate them from untested items. Use the impact on key user tasks to decide what should be fixed first.

Try These Prompts

Review a single form
Review this form for missing labels, unclear errors, and keyboard access. Explain each issue and list the manual checks I should run.
Triage a scanner report
Review these axe-core and pa11y findings. Group reports of the same issue, identify the affected controls, and say how to verify each issue manually.
Verify a checkout fix
Plan a keyboard and screen reader retest for this checkout modal. Check initial focus, Escape, focus return, and error announcements. Mark unavailable checks as untested.
Audit a release candidate
Audit the approved staging page using existing local tools. Ask before installing anything. Compare scanner findings with manual results and separate verified failures from checks that were not run.

Best Practices

  • Verify each automated finding in the browser before changing the interface.
  • Use native HTML controls where possible and retest behavior after each fix.
  • Record the browser, assistive technology, and checks that were not run.

Avoid

  • Claiming conformance after a clean automated scan.
  • Adding ARIA labels to custom controls without fixing their keyboard behavior.
  • Running an audit on an unapproved target or sharing raw reports without checking for sensitive content.

Frequently Asked Questions

Can this skill certify WCAG 2.2 AA conformance?
No. It helps find and fix barriers, but conformance requires evaluating all applicable criteria, including checks that automated tools cannot perform.
Which tools run in URL and file modes?
URL mode runs axe-core and pa11y. Local HTML file mode runs only pa11y.
What do I need to run the wrapper?
You need Bash, Node.js, and npm. The wrapper can fetch and run the pinned packages when they are not already available.
What happens when a scanner fails?
The wrapper reports an incomplete run and does not produce a combined violation total.
Will the skill verify screen reader behavior for me?
It provides a testing workflow. A check stays untested until someone verifies it with a named browser and screen reader.
Can a report contain sensitive page content?
Yes. JSON reports can contain HTML and text extracted from audited pages. Review and redact them before sharing.

Developer Details

Author

addyosmani

License

Unknown

Skillstore revision

r1

Version notice

The author did not declare a version.

Ref

b801e7ae7b71c5eca7822ed69f7a16176f3b6ed1

Maintenance freshness

11/20/2026

Usage

0 downloads ยท 0 views

File structure

๐Ÿ“ scripts/

๐Ÿ“„ audit-a11y.sh

๐Ÿ“„ SKILL.md