Community recipe · Tool workflows

Produce an evidence-bounded web accessibility audit

Source-backed example. Practical decision guidance is proposed and untested; this is not a full implementation guide.

The problem

Site owners need concrete accessibility findings while understanding which checks automation and available interaction evidence cannot actually establish.

Agent job

Confirm scope and target level, inspect available accessibility evidence, perform authorized checks and mark unsupported criteria as not tested.

Expected output

A criterion-level accessibility report with evidence and explicit testing limits, requiring specialist review before any conformance claim.

Where this framing comes from

Core problem: Editorial purpose framing
Agent job: Editorial purpose framing
Expected output: Editorial purpose framing
Relevant job roles: Editorial purpose framing
Benefit type: Editorial purpose framing

Editorial framing and outcome limits

Describes the task’s purpose and intended output; measured benefit, safe unsupervised operation and successful delivery are not established by this framing.

Editorial synthesis based on the linked source. This framing does not add evidence of successful delivery or results.

Case-specific decision guidance · untested

Case-specific decision guidance and first checks are original proposals. The linked source supports the attributed task; this added guidance has not been executed.

Read the attributed source evidence and original caveats below ↓

Why use this

A site owner can receive a clean automated accessibility report while important interaction and assistive-technology questions remain unanswered. Conversely, a detected issue may lack enough page-state evidence to identify its cause or reproduce it reliably.
An evidence-bounded audit plan could state which criteria have supporting observations and which need additional specialist testing. The owner could prioritize investigation without converting incomplete automation coverage into a whole-site conformance or legal-compliance claim.

Why an agent / simpler alternative

Useful judgment: Organize criterion-level evidence and identify unsupported conclusions or missing interaction states.
Simpler alternative: Use a standard automated scanner for a narrow preliminary check whose limitations are already understood.
Unsuitable when: The requested outcome is certification or an accessibility guarantee based only on automated observations.

Proposed inputs

Name
Audit scope brief
Owner
Site owner
Minimum scope
Authorized pages, states, target standard and output purpose
Quality check
Define excluded areas and safe interaction boundaries.
Name
Evidence sample
Owner
Accessibility reviewer
Minimum scope
Fictional observations, page-state labels and available manual evidence
Quality check
Separate observed results from checks not performed.

Access and setup — proposed or unknown

Proposed minimum: Offline evidence descriptors and a criterion-status proposal without browser interaction.
Compatibility unknowns: Unknown: current runtime, browser instrumentation, scanner coverage and assistive-technology access.
External actions requiring approval: Interactive testing scope, account access, test-data submission and external report publication.

Expected deliverable and review checks

Proposed artifact: Criterion-level evidence ledger with pass/fail candidates, untested areas and specialist questions.
Critical checks: Absent evidence remains not tested rather than becoming a pass.
Page scope and interaction state accompany each proposed finding.
Missing output evidence: No new workflow run or output inspection was performed for this overlay; use the canonical source-outcome references.

Failure signals and proposed responses

Signal
A whole-site pass is inferred from one scanned page.
Why it matters
The conclusion exceeds the inspected scope.
Proposed response
Restrict the statement to available evidence and list uncovered pages and states.
Signal
A keyboard result is claimed without an interaction record.
Why it matters
The finding lacks evidence for the behavior asserted.
Proposed response
Mark the criterion not tested and request an authorized manual observation.

Effort and cost — unknown

Setup work: Define scope, evidence requirements and specialist review ownership.
Ongoing review: Resolve unsupported findings and inspect later manual results.
Unknowns: Unknown: tooling charges, site complexity and assistive-technology review effort.

First check — proposed, unexecuted

Status: proposed_unexecuted
Synthetic fixture: A fictional audit summary claims site-wide success from one page and reports a keyboard pass without evidence.
Acceptance checks: Reduce the claim to its actual evidence scope.
Mark the keyboard criterion untested.
Stop conditions: Live scanning, browser interaction, certification or legal interpretation is required.

Limits

This proposed evidence audit validates neither accessibility behavior nor compliance; unsupported manual criteria remain untested.

Tasks: Research & discovery Security & compliance Writing & content

Useful for: Software developer Designer Product manager

Editorial relevance from the title and source-supported actions; not evidence of the operator’s occupation.

Claude Code / CodexIndependent testing: No

Editorial candidate. This is source-backed reporting or a documented workflow. BotCurrent has not independently tested this task. A proposed Canadian adaptation is not evidence of a Canadian deployment or compliance certification.

Original source

https://github.com/masuP9/a11y-specialist-skills/blob/main/skills/auditing-wcag/SKILL.md ↗

Attributed to: masuP9

Community recipe

Original action type: tool_action
What the source supports: Confirm page scope and target level, inspect the accessibility tree, run available automated checks and keyboard/focus interactions, request missing manual evidence, and write criterion-level pass, fail, not-tested or not-applicable findings.
Source outcome limits: Recipe only. Automated results do not certify whole-site accessibility or legal compliance; unsupported assistive-technology checks must remain untested.
Evidence attribution caveat: This community recipe describes a procedure; it does not prove successful execution or independently reproduced results.
Source approval description: Confirm the audit scope, level, limitations and output format; request evidence for checks the agent cannot perform. Unknown checks remain not tested.
Named tools: Claude Code or Codex; Playwright; axe and custom checks; accessibility tree
Model or version: Unknown
Publication date: Unknown
Source last checked: 2026-10-14
Demonstrated in Canada: Unknown
Canadian availability: Unknown
Proposed Canadian adaptation: Proposed: include English/French user flows and confirm which accessibility obligations and intended assistive-technology environments apply to the Canadian organization before publishing findings.
Recommended approval boundaries: Obtain site-owner permission for interactive tests; restrict to safe test data and authorized pages; require accessibility-specialist and assistive-technology review before any conformance claim.