AgentSkillsCN

executive-summary

从 GitHub 对话或会议记录中生成正式的执行摘要。适用于生成适合领导层的摘要,提炼复杂对话或会议中的关键决策、备选方案、结果与下一步行动。支持 GitHub issues/PRs 与记录 URI(Zoom、Teams 等)。输出保存为 Executive Summaries/,按日期整理结构,同时将原始输入归档至 Transcripts/,并匹配命名。

SKILL.md
--- frontmatter
name: executive-summary
description: Create formal executive summaries from GitHub conversations or meeting transcripts. Use when generating leadership-ready summaries that distill key decisions, alternatives, outcomes, and next steps from complex conversations or meetings. Supports GitHub issues/PRs and transcript URIs (Zoom, Teams, etc.). Outputs are saved to Executive Summaries/ with date-organized structure, and source inputs are archived to Transcripts/ with matching naming.

Executive Summary Skill

Create formal, narrative-driven executive summaries for leadership and stakeholders. This skill handles two primary workflows: synthesizing GitHub conversations (issues, pull requests, discussions) and distilling meeting transcripts (Zoom, Teams, etc.) into concise, decision-focused summaries.

Related Skills

Use brain-operating-system skill for:

  • Output directory structure and naming conventions (Executive Summaries/YYYY-MM-DD/, Transcripts/YYYY-MM-DD/)
  • Date folder creation patterns and file numbering

Use voice-and-tone skill for:

  • Narrative construction and paragraph patterns
  • First-person framing when appropriate
  • Crediting collaborators and showing impact

Use github-interaction skill for:

  • Fetching complete GitHub conversations (issues, PRs, discussions)
  • Comment and review retrieval patterns
  • Pagination handling

Core Principles

All executive summaries follow these unifying principles, regardless of source:

Narrative-Driven Prose

  • Structure content as dense, logically-connected paragraphs in formal, authoritative tone
  • Avoid bullet points, subheaders, or lists
  • Each paragraph builds on the previous, conveying a cohesive narrative of evolution from initial topic through decisions and next steps
  • Limit length to 3–5 structured paragraphs (GitHub summaries may run longer for complex decisions)

Impact & Decision Focus

  • Include only details that significantly influenced direction, decisions, or outcomes
  • Omit administrative commentary, routine pleasantries, subscription messages, procedural remarks, superficial technical minutiae (code diffs, exact timestamps), or automation events
  • Center on key debates, decisions, constraints, resolutions, and business/user impact
  • Clearly articulate alternatives explored, current status, next steps, and individual responsibilities

Contextual Linking

  • Every piece of cited information must link to its source
  • Attribute statements to individuals by name (no @ symbol on names themselves)
  • Links follow the statement or are integrated into sentences
  • Ensure readers can drill into source material for deeper context

Formal Tone & Authority

  • Use complete, well-structured sentences
  • Integrate all references and links seamlessly without extraneous formatting
  • Write for educated, time-constrained readers

Workflow: GitHub Conversations

For GitHub issues, pull requests, and discussions:

  1. Fetch the complete conversation using the github-interaction skill. This ensures you capture all comments, reviews, and state changes necessary for context.

  2. Identify the narrative arc: What was the initial problem/request? How did the conversation evolve? What decisions were made?

  3. Apply the rules from references/github-conversations.md, which details:

    • How to structure GitHub-specific summaries
    • Linking patterns for comments, events, and status changes
    • Handling alternative solutions and partial resolutions
    • Ignoring bot-generated events
  4. Archive the source conversation to Transcripts/YYYY-MM-DD/##.md (use today's date; sequential numbering within each date folder). Format as markdown with the issue/PR title, metadata, body, and all comments preserved.

  5. Save the summary to Executive Summaries/YYYY-MM-DD/##.md (use today's date; sequential numbering within each date folder)

Workflow: Meeting Transcripts

For Zoom, Teams, or other meeting transcripts:

  1. Fetch the transcript from the provided URI using HTTP or platform-specific tools

  2. Parse and prepare the transcript: Extract speaker attributions, timestamps, and key discussion points

  3. Apply the rules from references/transcript-summaries.md, which details:

    • Attribution patterns for named participants
    • How to reference shared documents or screens
    • Handling decisions and action items
    • Appropriate scope and constraints for transcript summaries
  4. Archive the source transcript to Transcripts/YYYY-MM-DD/##.md (use today's date; sequential numbering within each date folder). Preserve the full transcript text with speaker attributions.

  5. Save the summary to Executive Summaries/YYYY-MM-DD/##.md (use today's date; sequential numbering within each date folder)

Quick Reference: Which Workflow?

SourceUse This WorkflowNotes
GitHub issue, PR, discussion URLGitHub ConversationsFetch using github-interaction skill; apply GitHub-specific rules
Zoom/Teams transcript or recording URIMeeting TranscriptsFetch transcript; parse for speakers; apply transcript-specific rules
Email thread, Slack conversationGitHub Conversations (adapted)If available as a GitHub discussion or converted to one, use GitHub workflow; otherwise, treat as narrative text input

Tips for Quality Summaries

  • Start by understanding the arc: Skim the conversation or transcript to understand the trajectory before drafting
  • Prioritize decision impact: What changed as a result of this conversation? Lead with that
  • Use participant names strategically: Name decision-makers and key contributors; anonymize or skip minor commenters
  • Link judiciously but comprehensively: Every claim should be traceable; avoid standalone links
  • Edit for density: Remove connecting words, tighten sentences, but preserve clarity

Source Archival

Source materials (GitHub conversations, meeting transcripts) are archived to Transcripts/YYYY-MM-DD/##.md for future reference. These archives:

  • Follow the same naming pattern as executive summaries (YYYY-MM-DD/##.md) but are stored independently
  • Are not necessarily paired 1:1 with executive summaries - a transcript may exist without a summary, or vice versa
  • Can be referenced from anywhere in the brain: Meeting Notes, Weekly Notes, Daily Projects, or other documents via wikilinks

Naming Convention

Both folders use sequential numbering within each date:

FolderExample
Executive Summaries/2026-01-10/01.mdFirst summary of the day
Executive Summaries/2026-01-10/02.mdSecond summary of the day
Transcripts/2026-01-10/01.mdFirst transcript of the day
Transcripts/2026-01-10/02.mdSecond transcript of the day

Frontmatter

All output files must include YAML frontmatter. Generate a TID for each file:

bash
node ~/.copilot/skills/frontmatter-add/scripts/generate-tid.js

Executive summary frontmatter:

yaml
---
uid: <TID>
type: executive.summary
created: <ISO 8601>
tags: []
links:
  source: [<transcript TID if archived>]
  related: []
---

Transcript archive frontmatter:

yaml
---
uid: <TID>
type: transcript
created: <ISO 8601>
tags: []
links:
  related: []
---

The links.source field in the executive summary connects it to the archived transcript.

Archive Format

For GitHub conversations: Include title, URL, author, state, creation date, body, and all comments with author attribution and timestamps.

For meeting transcripts: Preserve the full transcript text with speaker names and timestamps as provided by the source platform.