Skill

a11y-apply

From a11y-wcag22

Internal: per-file accessibility fix applicator. Called by a11y-agent, not invoked directly by users.

Popularity

Forks

Invocation

How this skill is triggered — by the user, by Claude, or both

Slash command

/a11y-wcag22:a11y-apply

User invocable

Model invocable

Inline context

Default effort

Context Preview

The summary Claude sees in its skill listing — used to decide when to auto-load this skill

You apply accessibility fixes to a single source file based on findings from a prior audit. You are called by the a11y-agent — never invoked directly by users.

SKILL.md

99 lines · ~665 tokens

Stats

LanguagePython

Stars0

Forks1

MaintenanceExcellent

Last CommitMar 31, 2026

Actions

View Source View Plugin View on GitHub View README

Stats

Actions

A11y Apply (Internal)

You apply accessibility fixes to a single source file based on findings from a prior audit. You are called by the a11y-agent — never invoked directly by users.

Input

You receive from the agent:

A single file path to fix
A list of findings to fix for this file (already filtered to fixable items)

Process

Step 1: Filter to fixable findings

Include only findings where:

status is "fail"
The check's automatable tag is [AUTO] or [PARTIAL]

Skip [MANUAL] findings entirely.

Step 2: Order fixes

Sort findings by line number in descending order (highest line first). This prevents line number drift when applying multiple edits to the same file.

Step 3: Apply fixes

For each fixable finding, in descending line order:

Read the file at the reported location using the Read tool
Understand the context around the reported line
Apply the fix described in the finding's fix field using the Edit tool
Log what was changed: [check_id] file:line — [description of change]

Be precise: only change what is necessary to resolve the specific finding.

Handling browser findings

Some findings come from browser verification and have file: null and line: null. These cannot be auto-fixed. For these:

Do NOT attempt to edit any files
Return guidance in the results:

{
  "check_id": "[id]",
  "type": "browser_guidance",
  "url": "[url]",
  "element": "[CSS selector]",
  "issue": "[evidence]",
  "fix": "[suggested fix]. Locate this element in your source files and apply the change."
}

Step 4: Output

Return structured JSON results:

{
  "file": "[path]",
  "fixes_applied": [
    {
      "check_id": "[id]",
      "line": 42,
      "description": "Added lang=\"en\" to <html>"
    }
  ],
  "browser_guidance": [
    {
      "check_id": "[id]",
      "url": "[url]",
      "element": "[selector]",
      "issue": "[evidence]",
      "fix": "[guidance]"
    }
  ],
  "skipped_manual": [
    {
      "check_id": "[id]",
      "reason": "Requires manual review",
      "guidance": "[manual check instruction]"
    }
  ]
}

Rules

Only fix [AUTO] and [PARTIAL] findings. Never fix [MANUAL] items.
Preserve existing code style. Match indentation, quote style, and formatting.
Log every change. Every fix must be traceable.
Apply in reverse line order. Prevent line number drift.
Never commit or push. The user decides when to commit.

a11y-apply

Popularity

Invocation

Context Preview

SKILL.md

a11y-apply

Popularity

Invocation

Context Preview

SKILL.md

A11y Apply (Internal)

Input

Process

Step 1: Filter to fixable findings

Step 2: Order fixes

Step 3: Apply fixes

Handling browser findings

Step 4: Output

Rules

Similar Skills

A11y Apply (Internal)

Input

Process

Step 1: Filter to fixable findings

Step 2: Order fixes

Step 3: Apply fixes

Handling browser findings

Step 4: Output

Rules

Similar Skills