Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Write a decision-ready policy memo that frames an issue and recommends an option. Use when asked to write a policy memo, options paper, decision memo for a principal/minister/executive, or brief a decision-maker on a policy choice. Produces a tight memo: the issue, background, options with trade-offs, a clear recommendation, and implementation/risks — written for a busy decision-maker who reads the first paragraph.
.claude/skills/mohitagw15856-policy-memo/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 12% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 35% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 49% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 40% | 0% |
A policy memo exists to drive a decision, not to demonstrate research. The decision-maker reads the top and wants: what's the issue, what are my realistic options, what do you recommend, and what happens if I say yes. This skill writes that — BLUF (bottom line up front), honest options with trade-offs, and a defensible recommendation.
Ask for these only if they aren't already provided:
To / From / Date / Re — standard header.
Bottom line (BLUF) — 2–3 sentences: the issue, your recommended option, and the key reason. A busy reader should get the decision from this alone.
Issue — the precise question to be decided, and why it needs a decision now.
Background — only what's needed to decide (concise; detail goes to an annex). Facts, constraints, what's at stake.
Options — 2–4 realistic options (including status quo). For each: what it is, pros, cons, cost/feasibility, and who's affected. A comparison table helps:
| Option | Pros | Cons | Cost / feasibility | |---|---|---|---|
Recommendation — the option you recommend and why it best fits the goals and constraints. Be decisive; acknowledge the main trade-off you're accepting.
Implementation & risks — key steps, timeline, who does what, and the main risks + mitigations.
Next step / decision requested — exactly what you're asking the reader to approve.
Government & executive decision-memo practice (BLUF, options analysis, evidence-based recommendation, implementation).
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 27,943 | 27,407 | -2% | 1 | 1 | 0% | 3,426 | 3,711 | +8% | 0 | 0 | — |
case-02 | fail→pass | 26,276 | 24,084 | -8% | 1 | 1 | 0% | 3,176 | 3,542 | +12% | 0 | 0 | — |
case-03 | fail→fail | 23,437 | 31,171 | +33% | 1 | 1 | 0% | 2,804 | 4,053 | +45% | 0 | 0 | — |
case-04 | fail→pass | 24,187 | 26,319 | +9% | 1 | 1 | 0% | 2,921 | 3,939 | +35% | 0 | 0 | — |
case-05 | fail→pass | 23,662 | 30,776 | +30% | 1 | 1 | 0% | 3,168 | 4,717 | +49% | 0 | 0 | — |
case-06 | fail→fail | 20,330 | 27,806 | +37% | 1 | 1 | 0% | 2,314 | 3,765 | +63% | 0 | 0 | — |
case-07 | fail→fail | 22,701 | 29,452 | +30% | 1 | 1 | 0% | 2,610 | 3,799 | +46% | 0 | 0 | — |
case-08 | pass→pass | 24,801 | 27,681 | +12% | 1 | 1 | 0% | 2,414 | 3,445 | +43% | 0 | 0 | — |
case-09 | pass→pass | 29,530 | 27,237 | -8% | 1 | 1 | 0% | 3,157 | 4,086 | +29% | 0 | 0 | — |
case-10 | pass→pass | 23,600 | 26,407 | +12% | 1 | 1 | 0% | 2,529 | 3,982 | +57% | 0 | 0 | — |
case-11 | pass→pass | 18,063 | 25,628 | +42% | 1 | 1 | 0% | 1,864 | 3,271 | +75% | 0 | 0 | — |
case-12 | pass→fail | 68,861 | 54,992 | -20% | 1 | 1 | 0% | 8,223 | 8,170 | -1% | 0 | 0 | — |
case-13 | pass→pass | 21,451 | 26,325 | +23% | 1 | 1 | 0% | 2,048 | 3,336 | +63% | 0 | 0 | — |
case-14 | pass→pass | 25,749 | 20,817 | -19% | 1 | 1 | 0% | 2,989 | 2,950 | -1% | 0 | 0 | — |
case-15 | fail→pass | 29,002 | 26,314 | -9% | 1 | 1 | 0% | 2,742 | 3,842 | +40% | 0 | 0 | — |
case-16 | pass→pass | 27,286 | 31,342 | +15% | 1 | 1 | 0% | 3,418 | 4,212 | +23% | 0 | 0 | — |
case-17 | pass→pass | 14,226 | 15,154 | +7% | 1 | 1 | 0% | 2,089 | 2,577 | +23% | 0 | 0 | — |
case-18 | pass→pass | 19,823 | 20,682 | +4% | 1 | 1 | 0% | 2,683 | 3,630 | +35% | 0 | 0 | — |
case-19 | fail→fail | 18,224 | 20,101 | +10% | 1 | 1 | 0% | 2,437 | 2,881 | +18% | 0 | 0 | — |
case-20 | pass→pass | 16,922 | 16,895 | -0% | 1 | 1 | 0% | 2,304 | 3,221 | +40% | 0 | 0 | — |
case-21 | pass→pass | 24,127 | 37,870 | +57% | 1 | 1 | 0% | 3,088 | 3,617 | +17% | 0 | 0 | — |
case-22 | pass→pass | 21,266 | 18,365 | -14% | 1 | 1 | 0% | 3,067 | 3,487 | +14% | 0 | 0 | — |
case-23 | fail→pass | 17,602 | 19,120 | +9% | 1 | 1 | 0% | 2,596 | 3,663 | +41% | 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 +22 percentage points is the difference between those two pass rates over the 23 comparable cases. 1 case got worse with the skill loaded, and it is 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.