skills.rest
popup-studio-ai
popup-studio-aiOFFICIAL
Explore all skills by this author →
GitHub

pdca-watch

Renders a live 30-second dashboard of PDCA phase, match rate, and token usage from state files.

594|152|Updated Jan 9, 2026
Search Tags:#token usage#Workflow#pdca#Dashboard#live status#Monitoring#Claude code
One-click install
npx skills add https://github.com/popup-studio-ai/bkit-claude-code --skill pdca-watch
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: pdca-watch
Source: https://github.com/popup-studio-ai/bkit-claude-code/tree/main/skills/pdca-watch
Command: npx skills add https://github.com/popup-studio-ai/bkit-claude-code --skill pdca-watch

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

During long PDCA development cycles, you have no visibility into what the AI workflow is doing — which phase it is in, how close the code matches the design spec, or how many tokens it has consumed. This Skill turns that hidden state into a live, auto-refreshing dashboard.

Core Features & Use Cases

  • Live Tick Dashboard: Renders a fixed-width panel every 30 seconds via Claude Code's /loop, showing phase, match rate, iteration count, token totals, and estimated cost.
  • Auto Feature Resolution: Detects the active feature from pdca-status.json when no argument is given, or watches a named feature explicitly.
  • Safe Fallbacks: On Claude Code versions without /loop support, it renders a single snapshot with a warning instead of failing.
  • Use Case: While /pdca team runs a multi-agent sprint in the background, run the watch dashboard in parallel to monitor match-rate progress and token spend, and stop early if cost or drift exceeds expectations.

Quick Start

Ask the AI to start the live PDCA watch dashboard for the currently active feature.

Frequently Asked Questions about pdca-watch

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I monitor PDCA progress in real time?▼

Run the watch dashboard with no arguments to auto-detect the active feature, or pass a feature name explicitly. It renders phase, match rate, iteration count, and token usage every 30 seconds until the feature completes or 60 ticks elapse.

How do I track token usage and cost during a PDCA cycle?▼

The dashboard reads the last 50 lines of token-ledger.ndjson each tick and estimates cost using Sonnet-class rates of $3 per million input tokens and $15 per million output tokens.

Does the live dashboard work on older Claude Code versions?▼

Native 30-second looping requires Claude Code v2.1.71 or later with /loop support. On older or unknown versions it performs a single render with a warning instead of looping, as a fail-safe.

What happens if no PDCA feature is active?▼

If no feature argument is given and pdca-status.json has no primaryFeature or activeFeatures entry, the dashboard prints a message telling you to start one with /pdca pm and exits.

Does the watch dashboard modify any state files?▼

No. It is strictly read-only on pdca-status.json and token-ledger.ndjson, spawns no subprocesses, makes no network calls, and bounds the ledger tail to 200 lines regardless of input.

Related Skills

Explore More Software Engineering Skills→
solitasrohsolitasroh
2

pdca-batch

Coordinate up to three parallel PDCA cycles via the /pdca-batch workflow.

Community
Advanced
cuiosscuioss
4

phase-2-refine

Iteratively evaluate and refine plan requests until a confidence threshold is reached.

Official
Advanced
Euraika-LabsEuraika-Labs
14

sprint-status

Parse sprint plan and story files to generate a markdown progress report.

Official
Advanced
hendrax5hendrax5

pdca-batch

Manage multiple PDCA features with batch planning and queueing.

Community
Advanced