executive-summary

TL;DR — Transforms complex, multi-page content into crisp one-page executive summaries that surface decisions, metrics, and actions for time-constrained senior leaders.

---

When to Activate

Invoke this skill when any of these conditions are met:

Signal	Examples
Summarization request	"Write an executive summary of this report", "Resumen ejecutivo"
C-level preparation	"Summarize this for the board", "Brief the CEO on project status"
Decision framing	"Distill the options into a one-pager for the leadership team"
Document condensation	"Turn this 40-page analysis into something a VP will read"

Do not activate for meeting minutes (use meeting-notes), sprint reports (use sprint-report), or detailed technical documentation where brevity would compromise safety-critical information.

---

1. Source Material Analysis

Before writing, analyze the input material:

• Source type — full report, presentation deck, meeting transcript, data dump, or verbal briefing from the user.
• Length and density — estimate the compression ratio needed to reach one page (roughly 400-600 words).
• Audience — CEO, board, investors, department heads, or cross-functional leadership. Each audience has different priorities.
• Decision context — is this informational ("here is what happened"), decisional ("choose option A or B"), or action-oriented ("approve this plan")?
• Sensitivity level — confidential, internal, or shareable. Affects language and detail granularity.

If the source material is ambiguous or incomplete, state assumptions with [INFERRED] tags and flag gaps with [OPEN].

---

2. Executive Summary Framework

The skill uses the BLUF (Bottom Line Up Front) structure optimized for executive attention patterns:

2.1 Opening Statement (2-3 sentences)

The single most important takeaway. Answers the question: "If the reader stops here, what must they know?" This is the BLUF — state the conclusion, recommendation, or status first.

2.2 Context Block (3-4 sentences)

Minimal background needed to understand the opening statement. Assume the reader has strategic context but not operational detail. Reference time period, scope, and triggering event.

2.3 Key Metrics Table

Maximum 5-7 metrics in a compact table. Each metric includes: Metric Name, Current Value, Target/Benchmark, Status (on-track, at-risk, off-track). Only include metrics that directly support the narrative.

2.4 Decision Points or Findings (3-5 bullets)

If decisional: frame each option with pros, cons, and recommendation. Bold the recommended option. If informational: list the top findings in order of strategic importance. If action-oriented: list the requested approvals or actions.

2.5 Risks and Concerns (2-3 bullets)

Only the risks that would change the executive's decision or escalate their attention. Each risk includes probability indicator and mitigation status.

2.6 Recommended Actions (2-4 bullets)

Concrete, time-bound next steps with assigned owners. Framed as requests: "Approve X by date Y" or "Direct team Z to begin phase 2."

2.7 Appendix Pointer

Single line referencing the full source document for readers who want depth. Format: "Full analysis: [document name], [page count] pages, [date]."

---

3. Writing Principles

These principles govern the output quality:

• One page maximum — 400-600 words. If printed, it fits on a single A4/Letter page with standard margins.
• No jargon without context — define acronyms on first use or avoid them entirely.
• Active voice — "Revenue declined 12%" not "A 12% decline in revenue was observed."
• Quantify everything — replace qualitative assessments with numbers. "Significantly delayed" becomes "17 days behind schedule."
• Decision-ready — every summary should answer: "What do I need to know?" and "What do I need to do?"
• Evidence-tagged — all claims carry [EXPLICIT], [INFERRED], or [OPEN] tags.

---

4. Output Formatting

• Markdown mode (default): clean headers, a single compact table, bullet lists, and bold emphasis for key figures and recommendations.
• HTML mode: self-contained single-page HTML with inline CSS. Executive-friendly typography (serif headers, generous whitespace, status color indicators). Print-optimized with @media print rules.
• Slide mode: when the user says "for a slide" or "presentation format", produce content structured for a single presentation slide — title, 3-4 bullets, one key metric callout.
• Footer: Executive Summary | {source document} | {date} | Generated by executive-summary skill.

---

Trade-off Matrix

Dimension	Priority	Rationale
Clarity	HIGHEST	Ambiguity is the enemy of executive communication
Brevity	HIGH	One page is a hard constraint, not a suggestion
Accuracy	HIGH	Misrepresented data destroys trust instantly
Completeness	MEDIUM	Deliberately sacrificed for brevity; point to the full document
Visual polish	MEDIUM	Clean formatting aids scanning but substance wins

---

Assumptions & Limits

1. The skill summarizes content the user provides — it does not generate original analysis or fabricate data. [EXPLICIT]
2. Metrics are reproduced from source material. If no metrics exist, the skill highlights their absence. [EXPLICIT]
3. Recommended actions reflect the source material's conclusions. If the source has no recommendations, the skill flags this as an [OPEN] item. [EXPLICIT]
4. The skill assumes the audience is English-speaking C-level leadership unless told otherwise. [INFERRED]
5. For Spanish-language output ("resumen ejecutivo"), the skill produces the summary in Spanish while maintaining the same structural framework. [EXPLICIT]
6. Compression necessarily loses nuance — the appendix pointer is mandatory to preserve access to detail. [EXPLICIT]

---

Edge Cases

1. Source material is already short (under 500 words) — do not artificially pad. Restructure into the BLUF framework, add metrics table if data supports it, and note that the source was already concise.
2. Contradictory information in source — do not silently resolve contradictions. Surface them in the Risks section: "Source data contains conflicting figures for Q2 revenue (p.12 vs p.34). Clarification required before board presentation." Tag as [OPEN].
3. Highly technical source for non-technical audience — strip all technical detail. Replace implementation specifics with business impact statements. "Migrated to microservices architecture" becomes "Completed infrastructure modernization, reducing deployment time by 60%."
4. Multiple source documents — synthesize across sources into a single coherent summary. Note source documents in the appendix pointer. Flag any inter-document inconsistencies.

---

Good vs Bad Example

Good output (excerpt):

## Executive Summary: Project Atlas Q1 Status

Project Atlas is on track to deliver Phase 1 by June 2026, with 73%
of milestone deliverables completed. Budget utilization is at 68%
against a 75% plan, indicating a favorable variance. One critical
risk requires board attention: the vendor contract for the data
platform expires April 30 with no renewal signed. [EXPLICIT]

| Metric | Current | Target | Status |
|--------|---------|--------|--------|
| Milestone completion | 73% | 75% | On track |
| Budget utilization | 68% | 75% | Favorable |
| Team attrition | 2 of 14 | 0 | At risk |
| Customer pilot signups | 12 | 10 | Ahead |

**Recommended Actions:**
1. **Approve vendor contract renewal** by April 15 to avoid
   platform access disruption. Owner: VP Engineering. [EXPLICIT]
2. Backfill two departed engineers by Sprint 8. Owner: HR. [EXPLICIT]

Bad output (excerpt):

## Summary
The project is going well overall. We have done a lot of work and
things are mostly on schedule. There are some risks but the team is
handling them. We recommend continuing as planned.

The good example leads with the bottom line, provides a scannable metrics table with status indicators, and closes with specific, time-bound actions and owners. The bad example is vague, unquantified, and gives no actionable information.

---

Validation Gate

Before delivering the executive summary, confirm every item:

• Summary fits within one page (400-600 words maximum)
• Opens with a BLUF statement (conclusion or recommendation first)
• Context block provides minimal necessary background without operational detail
• Key metrics table contains 3-7 metrics with current value, target, and status
• Decision points or findings are ordered by strategic importance
• Risks section includes only escalation-worthy items with probability indicators
• Recommended actions are concrete, time-bound, and have assigned owners
• Appendix pointer references the full source document
• All claims carry evidence tags (EXPLICIT, INFERRED, OPEN)
• Language uses active voice, avoids jargon, and quantifies assessments
• Output format matches user request (markdown, HTML, or slide)
• Output file written to disk at confirmed path

---

Reference Files

File	Purpose
references/bluf-template.md	BLUF framework template with section prompts
references/body-of-knowledge.md	Executive communication best practices and BLUF methodology
references/html-template.html	Print-optimized HTML template for executive summaries
evals/evals.json	Evaluation scenarios for summary quality, compression ratio, and accuracy