Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Transform a prioritised initiative list into a compelling strategic roadmap narrative. Use when asked to write a roadmap narrative, explain the product roadmap to non-technical stakeholders, connect roadmap items to company goals, or produce an exec-shareable roadmap story. Produces a themed narrative with strategic context, quarter progression arc, an executive summary, and a 'what's not on the roadmap' section.
.claude/skills/mohitagw15856-roadmap-narrative/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -22% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 1875% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 155% | 0% |
| case-04 | ✓→✗ | ▼ Worse | -1% | 0% |
| case-05 | ✓→✗ | ▼ Worse | 84% | 0% |
Convert a ranked list of product initiatives into a clear, strategic narrative that connects individual items to company goals and communicates a coherent product direction.
If a professional-brain (brain/) exists, ground in it instead of re-asking for what you already know:
knowledge/strategy.md (the direction the narrative must ladder to), priority decisions/, and feature entities/. Run python3 ../professional-brain/scripts/brain_query.py ./brain "<roadmap theme>" and carry each fact's provenance tag through.decisions/ and updating the relevant feature entities/, each provenance-tagged. Show them, get a yes, then write with ../professional-brain/scripts/brain_write.py … --commit (append-only, dry-run by default).You will often get a short brief (a few themes, an audience) without a full initiative list or OKRs. Always deliver the complete narrative anyway — do not stop to ask questions and do not leave bracketed placeholders like [Theme Name]. Where detail is missing, infer specific, realistic themes, initiatives, and metrics from the brief and the domain, and mark any inferred fact or number as (assumed — confirm). Fill every section with concrete content, not template brackets.
Strategic Context: 1 paragraph: market moment, key challenge, our response]
Repeat for each theme]
What's Not on the Roadmap (and Why): 2-3 items with rationale — shows strategic discipline, not just prioritisation]
Executive Summary (shareable): 3-4 sentences that could be shared in an all-hands or board update]
When the themes have a sequence or dates, also render the roadmap as a Mermaid Gantt chart so the shape of the plan is visible (it renders live in the playground; with real ISO dates it also exports to a calendar .ics). Use section per theme/quarter and mark key checkpoints as milestones.
mermaidgantt title Roadmap dateFormat YYYY-MM-DD section Theme 1 Initiative :2026-07-01, 30d Checkpoint :milestone, 2026-07-31, 0d section Theme 2 Initiative :2026-08-01, 45d
This skill ships with support files — use them when they are available:
references/now-next-later.md — Now/Next/Later Done Right: Commitment Gradients, Not Date Camouflage. Apply it while producing the output; it carries the calibration and judgment calls the method summary above compresses.templates/roadmap-onepager.md — a fill-in version of the deliverable with the quality gates inline. Offer it when the user wants to work the document themselves rather than have it generated.Score any output of this skill before handing it over; 32+ is ship-quality.
| Dimension | 0 | 5 | 10 | |---|---|---|---| | Thematic coherence | A feature list with dates relabelled as a "narrative"; no themes | Themes exist but one or more initiatives are orphaned without being flagged | Every initiative maps to a theme; anything that doesn't fit is explicitly flagged as a narrative gap, not smuggled in | | Causal progression | Quarters are a chronological listing with no connection between them | Sequence is stated but the "Q1 enables Q2 because…" reasoning is missing or asserted without evidence | Each period visibly enables the next, with the causal dependency argued in prose and reflected in the timeline | | Strategic candour | No "what's not on the roadmap" section | One exclusion with thin rationale ("not a priority right now") | ≥2 named exclusions with defensible strategic rationale — including at least one idea someone actually wants | | Executive repeatability | Jargon-heavy; no standalone summary; a CFO couldn't retell any of it | Summary exists but is too long, too technical, or reads as compression rather than a story | 3–4 sentence summary a non-technical stakeholder could repeat correctly after one reading, zero engineering jargon |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 31,642 | 35,526 | +12% | 1 | 1 | 0% | 4,164 | 6,977 | +68% | 0 | 0 | — |
case-07 | fail→fail | 10,903 | 50,988 | +368% | 1 | 1 | 0% | 707 | 9,166 | +1196% | 0 | 0 | — |
case-02 | fail→fail | 32,662 | 35,672 | +9% | 1 | 1 | 0% | 3,872 | 6,979 | +80% | 0 | 0 | — |
case-03 | fail→pass | 52,361 | 29,205 | -44% | 1 | 1 | 0% | 7,293 | 5,725 | -22% | 0 | 0 | — |
case-04 | pass→fail | 57,307 | 48,639 | -15% | 1 | 1 | 0% | 8,232 | 8,146 | -1% | 0 | 0 | — |
case-05 | pass→fail | 27,251 | 29,703 | +9% | 1 | 1 | 0% | 2,984 | 5,499 | +84% | 0 | 0 | — |
case-06 | pass→fail | 19,101 | 34,451 | +80% | 1 | 1 | 0% | 1,921 | 5,658 | +195% | 0 | 0 | — |
case-08 | fail→fail | 12,267 | 41,229 | +236% | 1 | 1 | 0% | 1,030 | 5,335 | +418% | 0 | 0 | — |
case-09 | fail→pass | 8,016 | 44,460 | +455% | 1 | 1 | 0% | 365 | 7,209 | +1875% | 0 | 0 | — |
case-10 | fail→fail | 35,618 | 23,035 | -35% | 1 | 1 | 0% | 6,239 | 4,490 | -28% | 0 | 0 | — |
case-11 | pass→pass | 19,977 | 25,280 | +27% | 1 | 1 | 0% | 1,904 | 4,632 | +143% | 0 | 0 | — |
case-12 | fail→pass | 25,708 | 31,356 | +22% | 1 | 1 | 0% | 2,599 | 6,638 | +155% | 0 | 0 | — |
case-13 | pass→pass | 24,824 | 26,436 | +6% | 1 | 1 | 0% | 2,935 | 6,030 | +105% | 0 | 0 | — |
case-14 | fail→fail | 26,763 | 21,558 | -19% | 1 | 1 | 0% | 2,822 | 5,309 | +88% | 0 | 0 | — |
case-15 | pass→pass | 23,786 | 27,573 | +16% | 1 | 1 | 0% | 3,474 | 4,968 | +43% | 0 | 0 | — |
case-16 | pass→pass | 13,719 | 27,638 | +101% | 1 | 1 | 0% | 1,808 | 5,184 | +187% | 0 | 0 | — |
case-17 | fail→fail | 27,527 | 53,008 | +93% | 1 | 1 | 0% | 3,206 | 5,725 | +79% | 0 | 0 | — |
case-18 | pass→pass | 18,932 | 20,090 | +6% | 1 | 1 | 0% | 2,015 | 3,968 | +97% | 0 | 0 | — |
case-19 | pass→pass | 20,154 | 18,502 | -8% | 1 | 1 | 0% | 3,131 | 3,958 | +26% | 0 | 0 | — |
case-20 | pass→pass | 26,094 | 34,223 | +31% | 1 | 1 | 0% | 2,567 | 5,897 | +130% | 0 | 0 | — |
case-21 | fail→fail | 27,485 | 24,656 | -10% | 1 | 1 | 0% | 2,917 | 4,757 | +63% | 0 | 0 | — |
case-22 | pass→pass | 26,210 | 36,269 | +38% | 1 | 1 | 0% | 3,287 | 6,697 | +104% | 0 | 0 | — |
case-23 | pass→pass | 19,323 | 32,010 | +66% | 1 | 1 | 0% | 2,714 | 5,664 | +109% | 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 0 percentage points is the difference between those two pass rates over the 23 comparable cases. 3 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.