Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Open Orbit briefing skill — selected by the Orbit pipeline when GitHub is the user's only connected connector, or when the user explicitly scopes their daily digest to GitHub. Pulls the past 24 hours of PRs, review requests, issues, CI runs, and merges from the user's authenticated GitHub connection and renders them in a layout that mirrors GitHub's native Notifications + PR-diff visual language. This skill should not be triggered manually — it is invoked by Orbit's daily-digest scheduler agains
.claude/skills/nexu-io-orbit-github/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 115% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 84% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 139% | 0% |
Single-connector Orbit template scoped to GitHub.
Step 1. Open and read the shipped example.html in this folder before writing any output. That file is the canonical design — your job is to reproduce it, not reinterpret it.
Step 2. Mirror the example's structure 1:1:
<script> block at the end (filter / hover / link injection)Step 3. You may refresh mock values (PR numbers, titles, times, CI commit messages) so they read as "today", but you must not invent extra UI: no extra rail entries, no extra notifications, no extra event types, no extra badges, no extra chrome ornaments. If something is not already present in example.html, it does not belong in your output.
The sections below are a reference for tokens and visual language — not a license to extend the page.
This skill ships with its own complete visual language baked into example.html (GitHub's Primer chrome). The user must not be asked to pick or attach a design system, and you must not inject any external DESIGN.md tokens into the output.
example.html.This is a hard constraint: the briefing must read as a real GitHub page, not as the user's brand.
page bg: #f6f8fa
card bg: #ffffff
nav bar: #24292f /* GitHub black header */
nav text: #ffffff
ink: #1f2328
muted: #59636e
border: #d0d7de
hairline: rgba(208,215,222,0.32)
state · open: #1a7f37
state · merged: #8250df
state · closed: #cf222e
state · draft: #6e7781
attention bg: #fff8c5 /* yellow review-request block */
attention border: #d4a72c
ci-fail bg: #ffebe9
ci-fail border: #cf222eType stack:
-apple-system, BlinkMacSystemFont, "Segoe UI", "Noto Sans", Helvetica, Arial, sans-serif#24292f), 60px tall.Left: octocat SVG logo (white, 32px) + search input (rgba(255,255,255,0.08) background, white placeholder ghosted). Right: + plus dropdown, notifications bell with red dot if unread > 0, round avatar.
Left: page breadcrumb Inbox · Daily Digest · May 6. Right: filter dropdown chips (Type ▾ Date ▾ Status ▾).
Inbox · Saved · Done · All then divider then Participating · Mentions · Review requests · Assigned · Comments. Active item: light gray pill background.
(bg #fff8c5, 1px border #d4a72c). Each row: avatar + repo path + PR title + reviewer-state row of small dots (✓ green / ⏳ yellow / ○ gray) + "X of Y reviewers" + age.
(border-color #cf222e, bg #ffebe9) with a ✗ red glyph, run name, branch name (mono), commit message, age.
green / closed red), title, repo path, age, label pills.
small merged purple pill or closed red pill.
2–3 lines of mock code in a 12px monospace block with red − / green + prefixed lines and #ffebe9 / #dafbe1 row tints.
Open Orbit · auto-generated 06:42 · GitHub only.
Foreground white, background by state color above.
bug, p1, frontend …): GitHub label rounded pill, eachwith its own arbitrary color. Use varied real-world label hues.
✓ ⏳ ○ glyphs onlyif you can keep them visually subtle.
| Don't | Do | |---|---| | Mix light and dark themes | Stay on the light Primer theme (#f6f8fa page bg, #ffffff cards) | | Use non-GitHub typography | Use -apple-system, BlinkMacSystemFont, "Segoe UI", "Noto Sans", Helvetica, Arial, sans-serif exclusively | | Render avatars as squares or rounded squares | Always circles, with overlap ≤ 6px for reviewer stacks | | Use shadows / gradients / glows on chrome | Flat surfaces; differentiate with #d0d7de 1px borders | | Use lorem ipsum | Write real-shaped GitHub copy: PR titles like feat: orbit briefing card, branches like chore/upgrade-deps, commit subjects under 72 chars | | Render a CI failure as a normal row | Wrap in a red-bordered card (#cf222e border, #ffebe9 bg) with a red ✗ glyph and run name | | Render a review request as a normal row | Sit it in the yellow attention block (#fff8c5 bg, #d4a72c border) with reviewer status dots row | | Use placeholder repo names like org/repo | Use nexu-io/open-design (this org's actual primary repo) | | Pluck arbitrary label colors | Use realistic dev-team hues — bug red, enhancement blue, documentation light blue, frontend purple |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-22 | pass→pass | 15,322 | 19,476 | +27% | 1 | 1 | 0% | 2,923 | 5,756 | +97% | 0 | 0 | — |
case-01 | fail→fail | 50,796 | 52,626 | +4% | 1 | 1 | 0% | 8,262 | 10,007 | +21% | 0 | 0 | — |
case-02 | fail→fail | 41,806 | 38,876 | -7% | 1 | 1 | 0% | 8,259 | 10,004 | +21% | 0 | 0 | — |
case-03 | fail→pass | 50,762 | 37,323 | -26% | 1 | 1 | 0% | 8,248 | 9,993 | +21% | 0 | 0 | — |
case-04 | fail→fail | 37,849 | 53,416 | +41% | 1 | 1 | 0% | 8,216 | 9,961 | +21% | 0 | 0 | — |
case-05 | fail→pass | 37,910 | 38,312 | +1% | 1 | 1 | 0% | 8,015 | 9,967 | +24% | 0 | 0 | — |
case-06 | fail→fail | 38,109 | 48,019 | +26% | 1 | 1 | 0% | 8,214 | 9,959 | +21% | 0 | 0 | — |
case-07 | fail→fail | 39,923 | 49,472 | +24% | 1 | 1 | 0% | 8,212 | 9,957 | +21% | 0 | 0 | — |
case-08 | fail→fail | 48,190 | 34,920 | -28% | 1 | 1 | 0% | 4,543 | 9,954 | +119% | 0 | 0 | — |
case-09 | fail→pass | 23,815 | 47,561 | +100% | 1 | 1 | 0% | 4,631 | 9,947 | +115% | 0 | 0 | — |
case-10 | fail→pass | 33,530 | 46,394 | +38% | 1 | 1 | 0% | 5,400 | 9,957 | +84% | 0 | 0 | — |
case-11 | fail→pass | 21,252 | 49,219 | +132% | 1 | 1 | 0% | 4,167 | 9,949 | +139% | 0 | 0 | — |
case-12 | fail→fail | 22,300 | 38,856 | +74% | 1 | 1 | 0% | 3,594 | 9,954 | +177% | 0 | 0 | — |
case-13 | fail→pass | 42,488 | 38,064 | -10% | 1 | 1 | 0% | 6,933 | 8,084 | +17% | 0 | 0 | — |
case-14 | fail→pass | 15,286 | 36,490 | +139% | 1 | 1 | 0% | 2,239 | 9,796 | +338% | 0 | 0 | — |
case-15 | fail→pass | 30,025 | 7,030 | -77% | 1 | 1 | 0% | 5,054 | 3,030 | -40% | 0 | 0 | — |
case-16 | fail→pass | 31,401 | 39,450 | +26% | 1 | 1 | 0% | 5,126 | 8,740 | +71% | 0 | 0 | — |
case-17 | fail→pass | 13,770 | 38,061 | +176% | 1 | 1 | 0% | 2,604 | 9,955 | +282% | 0 | 0 | — |
case-18 | pass→pass | 8,488 | 12,842 | +51% | 1 | 1 | 0% | 1,429 | 4,415 | +209% | 0 | 0 | — |
case-19 | pass→fail | 28,233 | 47,889 | +70% | 1 | 1 | 0% | 4,426 | 9,949 | +125% | 0 | 0 | — |
case-20 | fail→pass | 31,110 | 49,388 | +59% | 1 | 1 | 0% | 6,373 | 9,947 | +56% | 0 | 0 | — |
case-21 | fail→fail | 23,293 | 42,678 | +83% | 1 | 1 | 0% | 4,858 | 9,951 | +105% | 0 | 0 | — |
case-23 | pass→pass | 9,372 | 8,052 | -14% | 1 | 1 | 0% | 1,443 | 3,203 | +122% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +43 percentage points is the difference between those two pass rates over the 23 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.