| name | lean-dashboard |
| description | Build an interactive HTML dashboard tracking sustainability findings and remediation over time — severity breakdowns, effort-vs-impact matrix, per-audit drill-down, and remediation status. Use this skill whenever the user wants to visualize or track Lean Agentic AI Skills audit results, monitor remediation progress across sprints, or present findings interactively rather than as a static report or deck. Consumes one or more lean-findings.json files; never audits anything itself. |
Green Dashboard
Pure consumer skill. Input: one or more lean-findings.json files (multiple files from the same subject over time enable trend view; different subjects enable portfolio view). Output: a single self-contained HTML file.
Prerequisite (platform-aware): if your environment provides a frontend-design skill (e.g. Claude's at /mnt/skills/public/frontend-design/SKILL.md), read it first — it governs styling there. Elsewhere, apply the dashboard's own build rules below; they are self-contained.
Dashboard structure
- Header — subject(s), audit date(s), producer skills, and a methodology drawer (collapsible, but present — the honesty rule follows findings wherever they render).
- Summary tiles — counts by severity, counts by SCI term, quick-win count (effort=low ∧ severity≥medium). No invented scores: a "green score of 73" is exactly the fabricated single number the suite refuses to produce; severity counts tell the truth.
- Effort-vs-impact matrix — findings plotted effort (x) × severity (y), click-through to detail. The screenshot slide of dashboards.
- Findings table — sortable/filterable by severity, SCI term, effort, producer; each row expands to issue/evidence/fix/pattern-link.
- Remediation tracking — status per finding (open / in-progress / done / won't-fix), persisted so it survives reloads; won't-fix requires a reason field (a documented decision is a legitimate outcome).
- Trend view (when multiple audits of the same subject exist) — findings opened vs closed over time by severity. Deltas between audit dates are real, countable numbers — the one place the dashboard can legitimately show progress figures.
- What's already good — the
passed items, visible by default, not buried.
Build rules
- Single self-contained HTML file; embed findings data as a JSON constant; no external CDN dependencies at runtime so it works offline and shares as one file.
- The dashboard practices the suite's own web patterns: no images, system font stack, no autoplaying anything, lean total size — it should itself pass web-carbon-audit.
- Persist remediation status via the environment's supported storage; where unsupported, keep in-memory and provide export/import of status as JSON so progress isn't lost.
- Charts in inline SVG, not chart libraries, for the simple visuals this needs.
Cost signal rendering
Cost driver column. Where findings carry cost_signal, add a driver column to the findings table (driver name and observed value). Trend view can chart the counted driver over time — this is the one place quantitative deltas ARE legitimate, because the input was counted.
