| name | team-visual-a11y |
| description | Unified team skill for visual accessibility QA. OKLCH color contrast, typography readability, focus management, WCAG AA/AAA audit at rendered level. Uses team-worker agent architecture. Triggers on "team visual a11y", "accessibility audit", "visual a11y". |
| allowed-tools | spawn_agent(*), wait_agent(*), send_message(*), followup_task(*), close_agent(*), list_agents(*), report_agent_job_result(*), request_user_input(*), Read(*), Write(*), Edit(*), Bash(*), Glob(*), Grep(*), mcp__ccw-tools__read_file(*), mcp__ccw-tools__write_file(*), mcp__ccw-tools__edit_file(*), mcp__ccw-tools__team_msg(*), mcp__chrome-devtools__evaluate_script(*), mcp__chrome-devtools__take_screenshot(*), mcp__chrome-devtools__emulate(*), mcp__chrome-devtools__lighthouse_audit(*), mcp__chrome-devtools__navigate_page(*), mcp__chrome-devtools__resize_page(*) |
Team Visual Accessibility
Deep visual accessibility QA: OKLCH-based perceptual color contrast, typography readability at all viewports, focus-visible completeness, WCAG AA/AAA audit at rendered level. Built on team-worker agent architecture -- all worker roles share a single agent definition with role-specific Phase 2-4 loaded from roles/<role>/role.md.
Architecture
Skill(skill="team-visual-a11y", args="task description")
|
SKILL.md (this file) = Router
|
+--------------+--------------+
| |
no --role flag --role <name>
| |
Coordinator Worker
roles/coordinator/role.md roles/<name>/role.md
|
+-- analyze -> dispatch -> spawn workers -> STOP
|
+-------+-------+-------+
v v v |
[3 auditors spawn in PARALLEL] |
color-auditor typo-auditor focus-auditor
| | |
+---+---+---+---+
v
remediation-planner
|
v
fix-implementer (inner_loop)
|
v
[re-audit: color + focus in PARALLEL]
Role Registry
Role Router
Parse $ARGUMENTS:
- Has
--role <name> -> Read roles/<name>/role.md, execute Phase 2-4
- No
--role -> roles/coordinator/role.md, execute entry router
Delegation Lock
Coordinator is a PURE ORCHESTRATOR. It coordinates, it does NOT do.
Before calling ANY tool, apply this check:
| Tool Call | Verdict | Reason |
|---|
spawn_agent, wait_agent, close_agent, send_message, followup_task | ALLOWED | Orchestration |
list_agents | ALLOWED | Agent health check |
request_user_input | ALLOWED | User interaction |
mcp__ccw-tools__team_msg | ALLOWED | Message bus |
Read/Write on .workflow/.team/ files | ALLOWED | Session state |
Read on roles/, commands/, specs/ | ALLOWED | Loading own instructions |
Read/Grep/Glob on project source code | BLOCKED | Delegate to worker |
Edit on any file outside .workflow/ | BLOCKED | Delegate to worker |
Bash("ccw cli ...") | BLOCKED | Only workers call CLI |
Bash running build/test/lint commands | BLOCKED | Delegate to worker |
If a tool call is BLOCKED: STOP. Create a task, spawn a worker.
No exceptions for "simple" tasks. Even a single-file read-and-report MUST go through spawn_agent.
Shared Constants
- Session prefix:
VA
- Session path:
.workflow/.team/VA-<date>-<slug>/
- team_name:
visual-a11y
- CLI tools:
ccw cli --mode analysis (read-only), ccw cli --mode write (modifications)
- Message bus:
mcp__ccw-tools__team_msg(session_id=<session-id>, ...)
- Max GC rounds: 2
Worker Spawn Template
Coordinator spawns workers using this template:
spawn_agent({
agent_type: "team_worker",
task_name: "<task-id>",
fork_turns: "none",
message: `## Role Assignment
role: <role>
role_spec: <skill_root>/roles/<role>/role.md
session: <session-folder>
session_id: <session-id>
requirement: <task-description>
inner_loop: <true|false>
Read role_spec file (<skill_root>/roles/<role>/role.md) to load Phase 2-4 domain instructions.
## Task Context
task_id: <task-id>
title: <task-title>
description: <task-description>
pipeline_phase: <pipeline-phase>
## Upstream Context
<prev_context>`
})
Parallel Fan-in Spawn (3 Auditors)
The 3 auditors run in parallel. Spawn all 3, then wait for all 3:
spawn_agent({
agent_type: "team_worker",
task_name: "COLOR-001",
fork_turns: "none",
message: `## Role Assignment
role: color-auditor
role_spec: ${skillRoot}/roles/color-auditor/role.md
session: ${sessionFolder}
session_id: ${sessionId}
requirement: ${colorTaskDescription}
inner_loop: false
Read role_spec file to load Phase 2-4 domain instructions.
## Task Context
task_id: COLOR-001
title: OKLCH Color Contrast Audit
description: ${colorTaskDescription}
pipeline_phase: audit`
})
spawn_agent({
agent_type: "team_worker",
task_name: "TYPO-001",
fork_turns: "none",
message: `## Role Assignment
role: typo-auditor
role_spec: ${skillRoot}/roles/typo-auditor/role.md
session: ${sessionFolder}
session_id: ${sessionId}
requirement: ${typoTaskDescription}
inner_loop: false
Read role_spec file to load Phase 2-4 domain instructions.
## Task Context
task_id: TYPO-001
title: Typography Readability Audit
description: ${typoTaskDescription}
pipeline_phase: audit`
})
spawn_agent({
agent_type: "team_worker",
task_name: "FOCUS-001",
fork_turns: "none",
message: `## Role Assignment
role: focus-auditor
role_spec: ${skillRoot}/roles/focus-auditor/role.md
session: ${sessionFolder}
session_id: ${sessionId}
requirement: ${focusTaskDescription}
inner_loop: false
Read role_spec file to load Phase 2-4 domain instructions.
## Task Context
task_id: FOCUS-001
title: Focus & Keyboard Accessibility Audit
description: ${focusTaskDescription}
pipeline_phase: audit`
})
wait_agent({ timeout_ms: 1800000 })
close_agent({ target: "COLOR-001" })
close_agent({ target: "TYPO-001" })
close_agent({ target: "FOCUS-001" })
After spawning, use wait_agent({ timeout_ms: 1800000 }) to collect results. If result.timed_out, send STATUS_CHECK via followup_task (wait 3 min), then FINALIZE with interrupt (wait 3 min), then mark timed_out and close agents. Use close_agent({ target }) each worker.
Model Selection Guide
Visual accessibility is a precision pipeline where auditors need thorough analysis and fix-implementer needs careful code changes.
| Role | reasoning_effort | Rationale |
|---|
| color-auditor | high | OKLCH calculations, contrast ratio precision |
| typo-auditor | high | Multi-breakpoint analysis, clamp() validation |
| focus-auditor | high | ARIA patterns, keyboard navigation completeness |
| remediation-planner | high | Synthesize 3 audit reports into actionable plan |
| fix-implementer | medium | Implementation follows defined remediation plan |
Audit-to-Remediation Context Flow
All 3 audit findings must reach remediation-planner via coordinator's upstream context:
// After COLOR-001 + TYPO-001 + FOCUS-001 all complete, coordinator sends findings to planner
spawn_agent({
agent_type: "team_worker",
task_name: "REMED-001",
fork_turns: "none",
message: `## Upstream Context
Color audit: <session>/audits/color/color-audit-001.md
Typography audit: <session>/audits/typography/typo-audit-001.md
Focus audit: <session>/audits/focus/focus-audit-001.md`
})
User Commands
| Command | Action |
|---|
check / status | View execution status graph |
resume / continue | Advance to next step |
Specs Reference
Session Directory
.workflow/.team/VA-<date>-<slug>/
+-- .msg/
| +-- messages.jsonl # Team message bus
| +-- meta.json # Pipeline config + GC state
+-- audits/
| +-- color/ # Color auditor output
| | +-- color-audit-001.md
| +-- typography/ # Typography auditor output
| | +-- typo-audit-001.md
| +-- focus/ # Focus auditor output
| +-- focus-audit-001.md
+-- remediation/ # Remediation planner output
| +-- remediation-plan.md
+-- fixes/ # Fix implementer output
| +-- fix-summary-001.md
+-- re-audit/ # Re-audit output (GC loop)
| +-- color-audit-002.md
| +-- focus-audit-002.md
+-- evidence/ # Screenshots, traces
v4 Agent Coordination
Message Semantics
| Intent | API | Example |
|---|
| Queue supplementary info (don't interrupt) | send_message | Send audit findings to running remediation-planner |
| Assign new work from reviewed plan | followup_task | Assign FIX task after remediation plan ready |
| Check running agents | list_agents | Verify agent health during resume |
Agent Health Check
Use list_agents({}) in handleResume and handleComplete:
// Reconcile session state with actual running agents
const running = list_agents({})
// Compare with tasks.json active tasks
// Reset orphaned tasks (in_progress but agent gone) to pending
Named Agent Targeting
Workers are spawned with task_name: "<task-id>" enabling direct addressing:
send_message({ target: "REMED-001", message: "..." }) -- send additional audit findings to remediation-plannerfollowup_task({ target: "FIX-001", message: "..." }) -- assign implementation from remediation planclose_agent({ target: "COLOR-001" }) -- cleanup after color audit
Error Handling
| Scenario | Resolution |
|---|
| Unknown command | Error with available command list |
| Role not found | Error with role registry |
| Session corruption | Attempt recovery, fallback to manual |
| Fast-advance conflict | Coordinator reconciles on next callback |
| Completion action fails | Default to Keep Active |
| GC loop stuck > 2 rounds | Escalate to user: accept / retry / terminate |
| Chrome DevTools unavailable | Degrade to static analysis only |
