agentleFS
EtanHeyEtanHey/golems/skills/golem-powers/t3-visuals/SKILL.md

t3-visuals

Skill0 starsChanged 16 days agoApache-2.0View on GitHub

Show/visualize/chart/diagram/compare/status board in T3 with compact interactive HTML. Not standalone dashboards.

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 4 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use for inline T3 html_render visuals, across orc, coach and leads.

    line 8
  2. Very specific:

    Keep short identity, unit, population, period and denominator labels.

    line 30
  3. Distinctive:

    No prose blocks inside or below the visual, expanded documents or explanatory lists.

    line 32
  4. Distinctive:

    Put omitted per-item meaning/precision in immediate hover AND keyboard-focus tooltips: one concise visible line.

    line 33
  5. Very specific:

    Use custom tooltips, no native-title delay or entrance animation.

    line 34
  6. Distinctive:

    Use light/dark-safe foreground/background pairs.

    line 38
  7. Very specific:

    Add linked hover/drill-down only when useful.

    line 48
  8. Very specific:

    Keep controls local and working (back/reset/replay when needed).

    line 49
  9. Very specific:

    Do first/middle/edge hover and focus show one normal-font line immediately, within bounds?

    line 55
  10. Distinctive:

    Do desktop/390px and light/dark preserve labels, scale and working controls?

    line 57

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

T3 visuals

Use for inline T3 html_render visuals, across orc, coach and leads. For a standalone published dashboard, use /html-dashboard; keep its delivery workflow.

Choose the encoding

Pick the relationship the reader needs to see; honor a supplied visual/behavior reference.

QuestionForm
Which work is ready or blocked?Grouped lane marks or a stage matrix
What changed?Explicit Old/Before and New/After pairs, same scale and units
When, in what order, or for how long?Timeline with exact local ranges on hover
Which option wins on which criteria?Aligned decision scorecard; keep unknown/not-run cells
How did magnitude or shares change over time?Line or stacked area; dashed forecast, solid observed
Where are the hotspots?Weekday/hour heatmap with an ordered intensity legend
Where is the concentration, spread or hierarchy?Concentration curve, distribution or treemap respectively
How does work move or depend on another step?Directed flow; label each stage and handoff
Only one meaningful number?One large value with unit/scope; no invented trend

Compose

Title the finding; put a few stable summary values above the data only when useful. Give data the surface: hairlines/space, dense aligned rows, right-aligned exact values, sparse ticks. Keep short identity, unit, population, period and denominator labels; distinguish build/test/merge/install. Show missing/estimated results honestly; never turn silence into success or a count into a version. No prose blocks inside or below the visual, expanded documents or explanatory lists; link long narrative separately. Put omitted per-item meaning/precision in immediate hover AND keyboard-focus tooltips: one concise visible line. Use custom tooltips, no native-title delay or entrance animation; copy .tip,.tip::before,.tip::after{font-style:normal;white-space:nowrap;max-width:calc(100vw - 16px);transition:none;animation:none} and clamp position to the viewport. Anchor near the mark, clamp to the viewport; shorten meaningfully rather than wrap, clip or hide essential detail. Color every status row/mark: green verified/ready, amber waiting/active, red blocked/error, gray not started/out of release. Pair state color with text/shape and a nearby key; keep entity hues and magnitude scales stable across views. Use light/dark-safe foreground/background pairs; measure all text/badges at ≥4.5:1 after alpha compositing. Copy these opaque AA pairs; apply the surface to the visual root and the matching status class to each badge/mark:

:root{--t3v-fg:#17212b;--t3v-muted:#435064;--t3v-bg:#ffffff;--t3v-ready-fg:#166534;--t3v-ready-bg:#dcfce7;--t3v-wait-fg:#713f12;--t3v-wait-bg:#fef3c7;--t3v-error-fg:#991b1b;--t3v-error-bg:#fee2e2;--t3v-idle-fg:#4b5563;--t3v-idle-bg:#e5e7eb}
@media(prefers-color-scheme:dark){:root{--t3v-fg:#f8fafc;--t3v-muted:#cbd5e1;--t3v-bg:#111820;--t3v-ready-fg:#86efac;--t3v-ready-bg:#153526;--t3v-wait-fg:#fde68a;--t3v-wait-bg:#3c2d14;--t3v-error-fg:#fca5a5;--t3v-error-bg:#3d2024;--t3v-idle-fg:#cbd5e1;--t3v-idle-bg:#27303b}}
.t3v-surface{color:var(--t3v-fg);background:var(--t3v-bg)}.t3v-muted{color:var(--t3v-muted)}
.t3v-ready{color:var(--t3v-ready-fg);background:var(--t3v-ready-bg)}.t3v-wait{color:var(--t3v-wait-fg);background:var(--t3v-wait-bg)}.t3v-error{color:var(--t3v-error-fg);background:var(--t3v-error-bg)}.t3v-idle{color:var(--t3v-idle-fg);background:var(--t3v-idle-bg)}

Local clocks always include am/pm, including heatmap/axis labels and both range endpoints: new Date(t).toLocaleTimeString('en-US',{timeZone:'Asia/Jerusalem',hour:'numeric',minute:'2-digit',hour12:true}).toLowerCase() → "5:22 am"; add date when needed and a space before the zone, e.g. 5:22 am Asia/Jerusalem or 4:30 am–5:10 am Asia/Jerusalem; never display ISO/UTC/Z. Label selectively without tiny type; preserve identities and units at desktop and 390px, with hover for omitted precision. Add linked hover/drill-down only when useful; update scope, KPIs, map, ranking and breadcrumb together. Keep controls local and working (back/reset/replay when needed); motion must explain change and respect reduced motion.

Before rendering

  • Does the chosen form answer the actual question with the supplied data and scope?
  • Can one glance find the result, units and each item's state without reading a text block?
  • Do first/middle/edge hover and focus show one normal-font line immediately, within bounds?
  • Are all text/badge contrasts measured, Old/New labels explicit, clocks local, and unknowns visible?
  • Do desktop/390px and light/dark preserve labels, scale and working controls?

Pass the HTML to html_preview, then html_render; never start a server for an inline visual.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.