AccessSutra AI
Local accessibility evidence. Not a WCAG conformance claim.
Local accessibility task testing: an offline HTML scan, a human test plan, and sealed reports. npm [email protected] published 2026-11-21. The command alias is accesssutra. The npm README says there is no API key and no telemetry, and 0 runtime dependencies. It says a scan, an empty findings list, a keyboard trace, or an agent reply is not a WCAG conformance claim, a VPAT, legal advice, or a substitute for testing with people who use assistive technology. On 2026-11-22 the npm downloads API returned no row, so both counts are 0. GitHub showed 0 stars and 0 open issues, and its detected language was HTML because the handbook is in the repository. The CLI file is packages/cli/accesssutra.mjs. The handbook home said 73 passed, 1 skipped, 0 failed. The npm README said 74 run, 73 passed, 0 failed, 1 skipped, recorded 2026-11-21 on Node v24.15.0. This page did not re-run either. The name AccessSutra has not been trademark-searched. WCAG is a W3C name. axe-core is by Deque. Playwright is by Microsoft. No affiliation is claimed.
- Offline
- 18 rules
- Node 22+
- npm latest
- [email protected]
- GitHub version
- 0.2.1
- Licence
- Apache-2.0
- Language
- JavaScript
- Weekly downloads
- 0
- Overall downloads
- 0
- GitHub stars
- 0
Use it when
- 1
You want an offline scan of a local HTML file, with line, selector, and a SHA-256 of that file.
- 2
You want a human test plan for keyboard, screen reader, zoom, contrast, and cognitive review.
- 3
You want a sealed JSON, Markdown, and HTML report without an API key.
What it does, and what it does not
Read this before the guide. A limit here is a limit of the tool.
Does
- scan reads a local HTML file. The npm README lists 18 rules, AS-DOC-001 through AS-REFRESH-001, each mapped to a candidate WCAG 2.2 criterion, with line, selector, and SHA-256 of the source.
- manual writes a plan of 10 human checks. The npm README says the tool never marks a manual check passed by itself, and that reviewer text which looks like a credential is rejected.
- report writes JSON, Markdown, and HTML. The npm README says the HTML report supports light, dark, and print.
- browser is optional and uses Playwright plus axe-core. The npm README says it accepts a file URL or a loopback URL only, does not submit forms, and was run once on both fixtures.
- audit writes a sealed assurance report. The npm README says that report covers 21 checks, a 56-case gold corpus, and fuzzing of 2,213 hostile inputs. verify checks the seal. audit append and audit verify keep a local hash-chained ledger.
- serve is a read-only report API on loopback. install-ide copies instruction files for nine IDE hosts. The npm README says native activation in each host is not verified. Node.js 22 or newer.
Does not
- It does not claim WCAG conformance. The npm README says the output is not a VPAT, an Accessibility Conformance Report, or legal advice.
- The static scan is not a browser. The npm README says it does not apply CSS, does not run JavaScript, and does not test contrast, focus visibility, keyboard traps, screen reader output, zoom, or cognitive load.
- A placeholder is not treated as an accessible name. The npm README says axe-core treats a placeholder as a name, so the two tools can disagree on that rule.
- The 56-case gold corpus is written by the maintainer. The npm README says precision 100 percent and recall 100 percent show internal consistency, not an independent benchmark.
- A SHA-256 seal and the audit ledger detect edits. The npm README says they are not signatures and not timestamps.
- The handbook home said 73 passed, 1 skipped, and 0 failed. The npm README said 74 run, 73 passed, 0 failed, and 1 skipped. This page did not re-run them and did not re-run the sample assurance report. The name AccessSutra has not been trademark-searched.
Claims above are from the project pages and are attributed as stated there. They are not independently verified here.
How a run moves
From the AccessSutra AI npm README, version 0.2.1, published 2026-11-21. It reads files you pass in. It does not decide that WCAG is met.
-
1. Scan
Static HTML
18 rules with line, selector, and a hash of the file. The npm README says this is not a rendered page.
-
2. Plan
10 human checks
Keyboard, screen reader, zoom, contrast, and cognitive review. The tool does not mark those passed.
-
3. Browser
Optional
Playwright and axe-core on a file or loopback URL. The npm README says forms are not submitted.
-
4. Seal
SHA-256
JSON, Markdown, and HTML. The npm README says the hash detects edits. It is not a signature and not a certificate.
How to read the result
Wording is from the project page or README and is attributed as stated there. It is not independently verified here.
Commands and notes
-
npx [email protected] doctorlocal check. No network. -
npx [email protected] demo --out ./accesssutra-demoJSON, Markdown, and HTML evidence report. -
npx [email protected] scan --input page.html --out scan.json --fail-on seriousexit 3 when that gate trips. -
npx [email protected] audit --out ./assuranceself-check bundle. verify checks the seal. -
manual, report, browser, install-ide, and servelisted in the npm README command table. browser and serve stay on a file or loopback.
Documentation
Titles below are the HTML titles from the GitHub Pages site on 2026-11-22. The handbook pages share one meta description, so that sentence is on the documentation home card only. The assurance report has no meta description. test-run.tap is not HTML, so that card uses the anchor text from the documentation home. This website did not re-run that assurance report.
- Documentation · AccessSutra AI AccessSutra AI documentation: local-first, evidence-based accessibility testing, WCAG 2.2 candidate rules, human test plans and sealed audit reports.
- Executive summary · AccessSutra AI
- Functional specification · AccessSutra AI
- Architecture · AccessSutra AI
- Technical reference · AccessSutra AI
- User guide · AccessSutra AI
- Security and privacy · AccessSutra AI
- Rules and evidence model · AccessSutra AI
- Test report · AccessSutra AI
- Operations and release · AccessSutra AI
- Legal and governance · AccessSutra AI
- Roadmap and FAQ · AccessSutra AI
- Glossary · AccessSutra AI
- AccessSutra AI — Assurance report
- Raw test log (TAP)
Questions
Does AccessSutra make a site WCAG compliant?
No. The npm README says a scanner result, an empty findings list, a keyboard trace, or an agent reply is not a WCAG conformance statement, a VPAT, legal advice, or a substitute for testing essential tasks with people who use assistive technology. This page is not legal advice either.
Where is the handbook, and did this page re-run the tests?
https://shahvatsalofficial.github.io/accesssutra-ai/ returned HTTP 200 on 2026-11-22. Thirteen handbook pages share one meta description, so this page shows that sentence on the documentation home card only. samples/audit/assurance-report.html and test-results/test-run.tap also returned HTTP 200 and have no meta description. The handbook home said 73 passed, 1 skipped, and 0 failed. The npm README said 74 run, 73 passed, 0 failed, and 1 skipped. This page did not re-run either, and it did not re-run the sample assurance report.
Other projects
- VigilSutra AI 247 last week · 2,243 overall · 4 stars
- RachnaSutra AI 214 last week · 963 overall · 4 stars
- ArchSutra AI 197 last week · 943 overall · 4 stars
- PramanaSutra AI 199 last week · 780 overall · 4 stars
- EvalSutra AI 209 last week · 592 overall · 4 stars
- DocSutra AI 216 last week · 567 overall · 4 stars
- IncidentSutra AI 540 last week · 540 overall · 4 stars
- ContextSutra AI 209 last week · 562 overall · 4 stars
- MigraSutra AI 213 last week · 553 overall · 4 stars
- WattSutra AI 491 last week · 491 overall · 4 stars
- ProtocolSutra AI 210 last week · 570 overall · 4 stars
- EdgeSutra AI 362 last week · 362 overall · 4 stars
- ContribSutra AI 512 last week · 512 overall · 4 stars
- LocalSutra AI 523 last week · 523 overall · 4 stars
- LineageSutra AI 337 last week · 337 overall · 0 stars
- DeploySutra AI 301 last week · 301 overall · 0 stars
- GovernSutra AI 459 last week · 459 overall · 0 stars