Executive Summary Writer

Lead with the pyramid's peak, not its base: the recommendation goes in the first sentence, not the last paragraph after the reader has already lost patience. This skill condenses a document into the answer plus exactly the support a reader needs to trust it — nothing more.

---

What this skill needs

• The source document.
• The specific decision or question the summary needs to support, if known.

---

Step 1: State the bottom line first

Write the recommendation or main conclusion as the first sentence.

Completion criterion: the first sentence is the recommendation or conclusion — not context, background, or a scene-setting lead-in.

---

Step 2: Support it

Add the 3–5 points a reader would ask "why" or "how do you know" next, each one traceable to the source document.

Completion criterion: every supporting point traces to a specific part of the source document, and there are no more than five.

---

Step 3: Cut everything else

Remove anything that doesn't serve the recommendation or the immediate follow-up questions. A one-page summary that reads like a shorter version of the whole document has failed at this step.

Completion criterion: the summary fits on one page and every sentence serves either the recommendation or a supporting point from Step 2.

---

Output format

One page: bottom-line sentence, then 3–5 supporting points as short paragraphs or bullets, no separate "introduction" or "background" section.

---

Evidence base

• Minto, The Pyramid Principle — backs leading with the conclusion and structuring supporting points as answers to the reader's next question, the organising method this skill applies throughout.

Meeting Notes & Actions

A meeting produced value only if the decisions are clear and the actions have an owner — otherwise it was just a conversation. This skill turns raw notes or a transcript into exactly that: decisions separated from discussion, and an action list that never leaves an item unowned.

---

What this skill needs

• The transcript, recording notes, or rough scribbles from the meeting.

---

Step 1: Separate discussion from decision

For each topic, write a one-line summary of what was discussed and mark the decision explicitly — or "no decision" if none was reached.

Completion criterion: every topic has an explicit decision marker, including "none," not just a discussion summary a reader has to infer a decision from.

---

Step 2: Build the action list

Pull actions only from things actually agreed — not every idea floated in discussion. Each action needs an owner and a due date.

Completion criterion: every action has a named owner and due date, or is flagged as missing one.

---

Step 3: Flag what's missing

Where an action has no clear owner or date in the source material, flag it rather than assigning one by guess.

Completion criterion: no action in the output has an invented owner or date not present in the source.

---

Output format

Decisions section, then an Actions table (Action | Owner | Due), then an optional brief Discussion notes section for context.

---

Evidence base

• PMBOK Guide — backs separating decisions from discussion and tracking actions with a named owner and date as standard project-communication practice.
• Perlow et al., Stop the Meeting Madness (HBR) — backs unresolved, un-owned action items as a primary driver of meetings that don't actually move work forward, which is what Step 3's flag targets.

OKR Check-In

A confidence score with no reason attached is a guess dressed up as data. This skill's discipline is attaching a stated reason to every score, and naming the specific blocker behind anything below target — not a vague "behind schedule."

---

What this skill needs

• The objective and key results as set.
• Raw progress notes for the period.

---

Step 1: Score each key result

For each key result, state the current value against target and a confidence score (0.0–1.0, or red/amber/green) based on what the notes actually support — not an optimistic default.

Completion criterion: every key result has a stated current value and a confidence score with a one-line reason.

---

Step 2: Name blockers

Where confidence is below target, name the specific blocker rather than a generic "behind schedule" or "in progress."

Completion criterion: every below-target key result has a named, specific blocker, not a restatement of the low score.

---

Step 3: Note the path back, if given

If the owner's notes state what would need to change to get back on track, include it. Don't invent a recovery plan they didn't give.

Completion criterion: any recovery plan in the output was present in the source notes, not generated to fill the gap.

---

Output format

Per key result: Target | Current | Confidence (with reason) | Blocker (if below target) | Path back (if stated).

---

Gotchas

• A confidence score with no reason is a guess dressed as data — always attach the one-line reason.

---

Evidence base

• Doerr, Measure What Matters — backs the confidence-scored check-in (rather than percent-complete alone) as the OKR-review discipline, since it surfaces risk before the quarter ends rather than at the deadline.
• Google re:Work OKR guidance — backs naming specific blockers over generic "on track/behind" framing.

Performance Review Draft

Generic praise ("great team player") without a specific example behind it is the most common weak spot in review writing — it doesn't help the person improve and doesn't hold up if challenged. This skill ties every line to a concrete example, and leaves the rating decision entirely to the manager.

---

What this skill needs

• The competencies or goals the review is measured against.
• The manager's notes and examples for the period — specific incidents, not general impressions.

---

Step 1: Map examples to competencies

For each competency or goal, find the example(s) that evidence it. Where a competency has no example behind it, flag that rather than forcing a vague sentence to cover the gap.

Completion criterion: every competency either has a specific example or is explicitly marked "no evidence this period."

---

Step 2: Draft the narrative

Per competency: what was observed (specific, not generic praise or criticism), its impact, and — for a development area — a concrete suggestion for what to do differently.

Completion criterion: no sentence uses an evaluative adjective ("great," "needs improvement") without a specific example attached.

---

Step 3: Leave the rating to the manager

Do not assign or suggest a rating, grade, or score. Mark that field [Manager to complete] unless the manager has directly supplied it.

Completion criterion: no rating or grade appears in the draft unless the manager supplied it directly.

---

Output format

One section per competency: Observed / Impact / Suggestion (if a development area). Rating field left blank or marked for the manager.

---

Gotchas

• A competency with no concrete example behind it should be flagged, not papered over with generic language.
• The rating judgement always stays with the manager — this drafts the narrative that supports it, not the decision itself.

---

Evidence base

• SHRM guidance — backs evidence-based, competency-linked narrative over generic praise/criticism language, and backs keeping the rating decision a distinct step from the narrative.
• Gallup, State of the Global Workplace — backs specific, example-based feedback as what actually drives engagement, versus generic evaluative language.

Stakeholder Update Drafter

A stakeholder reading this email is asking one question first: is this on track, and do you need anything from me? This skill puts that answer in the first line, then backs it up — matching whatever voice has already been established, rather than writing in a generic corporate register.

---

What this skill needs

• Project notes or updates for the period.
• The recipient and their stake — decision-maker, informed party, sponsor.
• A voice sample, if the sender wants their own tone matched (not required).

---

Step 1: Match the voice

If a voice sample or established tone is already in context, apply it silently — don't mention it. Otherwise default to clear, warm, professional English without asking, unless the user offers a sample unprompted.

Completion criterion: the draft's tone is consistent with any established voice in context, or clear and professional by default.

---

Step 2: Lead with the headline

Open with the status in one line — on track, needs attention, or decision needed — then 2–4 supporting points.

Completion criterion: the first line states status; supporting points don't exceed four.

---

Step 3: State the ask

If anything is needed from the recipient, make it explicit rather than implied within a paragraph, and attach a date if one exists.

Completion criterion: any ask of the recipient is its own clearly marked line, not buried in supporting detail.

---

Output format

Headline line, 2–4 supporting points, explicit ask (if any) with a date, closing line.

---

Evidence base

• Minto, The Pyramid Principle — backs the headline-first structure this skill applies to a stakeholder audience.
• PMBOK Guide — backs distinguishing status communication aimed at a decision-maker (needs the ask stated up front) from a general project update with no action required.

Project Status Report

A status report exists to support a go/no-go decision at steering level — which only works if a red status is actually reported red. This skill assigns RAG ratings from the evidence in your notes, never a default green, and never softened to make the report read better.

---

What this skill needs

• Raw updates, notes, or a stand-up transcript for the period.
• The reporting period and, if the audience expects one, the template.

---

Step 1: Assign RAG status with a reason

For each workstream or milestone, assign red/amber/green based on what the notes actually support, and state the one-line reason behind anything not green.

Completion criterion: every RAG rating has a stated reason; no rating is a default with no evidence behind it.

---

Step 2: List milestones

State each milestone due this period as hit, missed, or at risk.

Completion criterion: every milestone due this period appears with one of the three states, not omitted because it's unresolved.

---

Step 3: List risks and blockers

For each, name the owner and the next step — not just a description of the problem.

Completion criterion: every red/amber item has a named next step and owner, or is flagged as needing one assigned.

---

Output format

RAG summary line, then Milestones table, then Risks/Blockers table (Risk | Owner | Next Step), then a short Next Period section.

---

Gotchas

• Don't soften a red status to amber to make the report read better — an unstated risk is exactly the failure mode this exists to prevent.

---

Evidence base

• PMBOK Guide — backs RAG status with a stated basis, not a vibe check, and risk-with-owner as standard reporting discipline.
• PRINCE2 — backs the milestone-plus-risk structure as the minimum a steering-level report needs to support a go/no-go decision.