Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Audience-specific tone and format guidance for leadership communications. Use when drafting any /report:* output to tune length, framing, and technical depth to the reader (board, audit committee, CEO, weekly CISO, regulator).
.claude/skills/grcengclub-exec-narrative-patterns/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-10 | ✗→✓ | ▲ Improved | 44% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 53% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-16 | ✓→✓ | ＝ Same ✓ | 19% | 0% |
Different readers want different things. The same findings should not land the same way in a board deck, a CISO 1:1, and a regulator response.
| Audience | Length budget | Opening line wants | Technical depth | What they skip | | --- | --- | --- | --- | --- | | Full board | 1 page max | Strategic posture one-liner | Low. No control IDs in body. | Operational detail, acronyms, tables with >5 rows | | Audit committee | 2 pages | Risk-and-assurance posture | Medium. Control families OK, not IDs. | Strategy monologue, vendor names unless load-bearing | | Risk committee | 2 pages | Residual risk movement | Medium-high. Framework names, risk scoring. | Program branding, tool selection detail | | CEO weekly | Half page | What changed, what's blocking, what's asked | Low. Translate everything. | Framework politics, tool comparisons | | CISO weekly | 1 page | What moved, what's blocked, what's next | High. Control IDs, tool names, owner names fine. | Over-explained context | | Regulator / auditor | As long as needed | Precise scope + evidence | Maximum. IDs, artifact paths, timestamps. | Narrative softening |
Board and audit committee
CEO weekly
CISO weekly
Regulator / auditor
YYYY-MM-DD. Periods are YYYY-Q# or YYYY-W##.If you cannot defend every sentence as "the audience needed this to make a decision," cut it.
If the document opens with context before the point, invert it.
If there are three open items and no ask, you are reporting, not communicating. Add the ask.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-16 | pass→pass | 20,496 | 18,061 | -12% | 1 | 1 | 0% | 3,040 | 3,621 | +19% | 0 | 0 | — |
case-03 | fail→fail | 18,043 | 17,333 | -4% | 1 | 1 | 0% | 2,749 | 3,500 | +27% | 0 | 0 | — |
case-01 | fail→fail | 14,664 | 19,504 | +33% | 1 | 1 | 0% | 2,199 | 2,512 | +14% | 0 | 0 | — |
case-02 | fail→fail | 13,711 | 9,372 | -32% | 1 | 1 | 0% | 1,898 | 2,001 | +5% | 0 | 0 | — |
case-15 | fail→fail | 9,216 | 8,963 | -3% | 1 | 1 | 0% | 1,400 | 2,044 | +46% | 0 | 0 | — |
case-04 | fail→fail | 12,192 | 11,632 | -5% | 1 | 1 | 0% | 1,902 | 2,407 | +27% | 0 | 0 | — |
case-05 | fail→fail | 20,679 | 18,625 | -10% | 1 | 1 | 0% | 3,214 | 3,462 | +8% | 0 | 0 | — |
case-06 | fail→fail | 17,737 | 14,596 | -18% | 1 | 1 | 0% | 2,579 | 2,763 | +7% | 0 | 0 | — |
case-07 | fail→fail | 11,051 | 8,951 | -19% | 1 | 1 | 0% | 1,543 | 1,979 | +28% | 0 | 0 | — |
case-08 | fail→fail | 19,035 | 24,015 | +26% | 1 | 1 | 0% | 2,799 | 4,266 | +52% | 0 | 0 | — |
case-09 | fail→fail | 8,645 | 6,875 | -20% | 1 | 1 | 0% | 1,229 | 1,610 | +31% | 0 | 0 | — |
case-10 | fail→pass | 17,815 | 21,453 | +20% | 1 | 1 | 0% | 2,694 | 3,890 | +44% | 0 | 0 | — |
case-11 | fail→pass | 13,072 | 13,500 | +3% | 1 | 1 | 0% | 1,832 | 2,808 | +53% | 0 | 0 | — |
case-12 | fail→pass | 7,684 | 5,461 | -29% | 1 | 1 | 0% | 1,124 | 1,478 | +31% | 0 | 0 | — |
case-13 | fail→pass | 19,335 | 9,957 | -49% | 1 | 1 | 0% | 3,039 | 2,132 | -30% | 0 | 0 | — |
case-14 | fail→fail | 15,437 | 14,569 | -6% | 1 | 1 | 0% | 2,259 | 3,014 | +33% | 0 | 0 | — |
case-17 | fail→fail | 21,103 | 12,593 | -40% | 1 | 1 | 0% | 3,071 | 2,532 | -18% | 0 | 0 | — |
case-18 | fail→fail | 17,499 | 20,293 | +16% | 1 | 1 | 0% | 2,558 | 3,759 | +47% | 0 | 0 | — |
case-19 | fail→fail | 10,360 | 7,511 | -28% | 1 | 1 | 0% | 1,465 | 1,680 | +15% | 0 | 0 | — |
case-20 | pass→pass | 12,982 | 10,599 | -18% | 1 | 1 | 0% | 2,440 | 2,459 | +1% | 0 | 0 | — |
case-21 | pass→pass | 25,219 | 25,174 | -0% | 1 | 1 | 0% | 4,028 | 4,715 | +17% | 0 | 0 | — |
case-22 | pass→pass | 12,230 | 10,559 | -14% | 1 | 1 | 0% | 2,012 | 2,335 | +16% | 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 +18 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.