Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Translate one internal roadmap into audience-appropriate formats with confidence-band discipline. Use when preparing a roadmap readout, tailoring it for an exec, customer, eng, or sales audience, or auditing over-promise risk.
.claude/skills/borghei-roadmap-communicator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 222% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 72% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 222% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 83% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 80% | 0% |
A skill focused on communicating the roadmap — different audiences need different formats and confidence levels. Distinct from product-strategist (which builds the strategy) and agile-product-owner (which manages sprint execution).
Before generating the roadmap communication, confirm these inputs. If any is unknown or vague, ASK — do not assume:
Stop rule: ask only the 2-3 that most change the output. If the user says "just draft it," proceed and list your assumptions at the top of the artifact.
roadmap_audience_translator.py with target audience.bashpython3 roadmap-communicator/scripts/roadmap_audience_translator.py \ --input roadmap.json --audience customer --format markdown
confidence_band_generator.py with team velocity history + estimation context.bashpython3 roadmap-communicator/scripts/confidence_band_generator.py \ --input items.json --format markdown
roadmap_diff_reporter.py to produce what-changed memo.bashpython3 roadmap-communicator/scripts/roadmap_diff_reporter.py \ --previous roadmap_q1.json --current roadmap_q2.json --format markdown
| Audience | Right format | Wrong format | |----------|--------------|--------------| | Board / exec | Themes + bets + KPIs (1 page) | Feature list | | Customers / public | What's new + what's next (themes; no dates) | Internal commit list | | Sales | Themes + competitive positioning + customer-ask coverage | Engineering jargon | | Engineering | Themes + quarter commitments + scoped detail | Vague aspirations | | Internal company | Themes + progress + asks | Confidential strategy | | Partner / integrator | API-relevant changes + breaking-change calendar | All-up roadmap |
Same roadmap; different formats. Don't send the engineering commit list to customers.
Apply per item:
| Band | Language | Audience expectation | |------|----------|----------------------| | Commit | "Will ship" with target date | Hold us to this | | Plan | "Plan to ship" with target window | Confident but conditional | | Aspire | "Investigating" / "Exploring" | Don't depend on this | | Strategic intent | "We believe X matters" | Direction, not deliverable |
Common errors:
A useful skeleton across audiences:
This protects confidence: the closer in time, the firmer the commitment.
Communicate at the right granularity:
Telling a customer "we're adding X in Q3" makes a commitment that may not be precise enough. Telling the team "we're going to improve collaboration somehow" is too vague.
references/roadmap-communication-patterns.md — format catalog + when to usereferences/audience-specific-formats.md — per-audience templatesreferences/now-next-later-and-themes.md — structural patternsproduct-team/product-strategist — strategy upstream of roadmapproduct-team/agile-product-owner — sprint-level executionproduct-team/product-manager-toolkit — broader PM toolingc-level-advisor/cpo-advisor — CPO partnershipc-level-advisor/ceo-advisor — CEO / board alignmentbusiness-growth/customer-success-manager — customer commsmarketing/ skills — external messaging alignment| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 5,927 | 6,911 | +17% | 1 | 1 | 0% | 856 | 2,754 | +222% | 0 | 0 | — |
case-02 | fail→pass | 14,466 | 12,835 | -11% | 1 | 1 | 0% | 2,127 | 3,662 | +72% | 0 | 0 | — |
case-03 | fail→fail | 5,024 | 6,554 | +30% | 1 | 1 | 0% | 845 | 2,645 | +213% | 0 | 0 | — |
case-04 | fail→fail | 29,717 | 32,777 | +10% | 1 | 1 | 0% | 4,698 | 6,621 | +41% | 0 | 0 | — |
case-05 | fail→fail | 7,483 | 10,772 | +44% | 1 | 1 | 0% | 1,273 | 3,531 | +177% | 0 | 0 | — |
case-06 | fail→pass | 5,763 | 9,611 | +67% | 1 | 1 | 0% | 959 | 3,091 | +222% | 0 | 0 | — |
case-07 | fail→pass | 13,233 | 14,159 | +7% | 1 | 1 | 0% | 2,111 | 3,862 | +83% | 0 | 0 | — |
case-08 | pass→pass | 10,338 | 14,500 | +40% | 1 | 1 | 0% | 1,608 | 3,649 | +127% | 0 | 0 | — |
case-09 | pass→pass | 13,216 | 13,257 | +0% | 1 | 1 | 0% | 2,021 | 3,842 | +90% | 0 | 0 | — |
case-10 | pass→pass | 6,652 | 10,318 | +55% | 1 | 1 | 0% | 1,217 | 3,332 | +174% | 0 | 0 | — |
case-11 | fail→pass | 12,640 | 10,578 | -16% | 1 | 1 | 0% | 1,887 | 3,405 | +80% | 0 | 0 | — |
case-12 | pass→pass | 11,528 | 11,303 | -2% | 1 | 1 | 0% | 1,641 | 3,393 | +107% | 0 | 0 | — |
case-13 | fail→pass | 8,155 | 8,200 | +1% | 1 | 1 | 0% | 1,292 | 2,895 | +124% | 0 | 0 | — |
case-14 | fail→pass | 7,500 | 2,555 | -66% | 1 | 1 | 0% | 1,201 | 2,184 | +82% | 0 | 0 | — |
case-15 | fail→pass | 9,840 | 3,510 | -64% | 1 | 1 | 0% | 1,453 | 2,247 | +55% | 0 | 0 | — |
case-16 | fail→pass | 5,527 | 2,784 | -50% | 1 | 1 | 0% | 815 | 2,081 | +155% | 0 | 0 | — |
case-17 | fail→pass | 9,029 | 7,516 | -17% | 1 | 1 | 0% | 1,398 | 2,750 | +97% | 0 | 0 | — |
case-18 | fail→pass | 13,972 | 10,415 | -25% | 1 | 1 | 0% | 2,229 | 3,339 | +50% | 0 | 0 | — |
case-19 | pass→pass | 12,795 | 9,909 | -23% | 1 | 1 | 0% | 2,020 | 3,245 | +61% | 0 | 0 | — |
case-20 | pass→pass | 12,872 | 13,615 | +6% | 1 | 1 | 0% | 1,882 | 3,798 | +102% | 0 | 0 | — |
case-21 | pass→pass | 10,255 | 8,082 | -21% | 1 | 1 | 0% | 1,550 | 2,872 | +85% | 0 | 0 | — |
case-22 | pass→pass | 8,795 | 10,789 | +23% | 1 | 1 | 0% | 1,404 | 3,293 | +135% | 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. 22 cases were attempted. The headline lift of +50 percentage points is the difference between those two pass rates over the 22 comparable cases.
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.