I Let AI Prepare Me for Every Meeting With My Manager.

---
name: daily-brief
description: Generate a comprehensive daily brief summarizing meetings, completed work, work in progress, events, and community insights. Use this skill when the user asks to generate or create their daily brief.
license: MIT
---

# Daily Brief Skill

## Purpose
When the user requests their daily brief, generate a comprehensive summary of the workday following the specified format.

## ⚠️ MANDATORY FIRST STEP
**BEFORE STARTING ANY BRIEF GENERATION:** Always follow the workflow exactly as specified. Do not use generic WorkIQ queries or alternative data sources.

## Trigger
- "Generate my daily brief"
- "Create daily brief for [date]"
- "What's my daily brief?"
- "Daily brief"
- "Generate my daily brief for [date]"
- And similar variations

## Date Handling
- Default to today's date (current_datetime) unless user explicitly specifies a different date
- Parse relative dates like "yesterday", "last Friday", etc.
- Confirm date with user if unclear
- **TIMEZONE**: All daily brief boundaries use **PDT (UTC-7, California timezone)**
- **WORKDAY BOUNDARY**: Your workday ends at **10PM PDT**. Include any meetings or completed work that occur up to 10PM PDT on that calendar day (e.g., an issue closed at 8PM on July 1 PDT is included in the July 1 brief, even though it's technically July 2 UTC)

## Data Collection Process

### 1. Meetings
**Tool**: WorkIQ `workiq-ask` - Calendar
- Use `workiq-ask` to query calendar for the specified date
- **Ask**: "For [DATE], list ONLY calendar events whose category is exactly `Required Meeting`. For every returned event, provide its exact categories, title, start/end time, organizer, attendees, and available notes. Do not infer the category from attendees, title, organizer, or whether it appears to be a work meeting. Exclude every event whose categories are missing or do not contain the exact `Required Meeting` label."
- Pass `timeZone: "America/Los_Angeles"` to get PDT-converted times automatically
- **Filter by: Category = "Required Meeting" ONLY** - include an event only when the WorkIQ response explicitly reports that its categories contain the exact `Required Meeting` label. Do not include an event based on attendees, organizer, title, transcript availability, or an inferred work-meeting classification.
- **Category verification gate**: Before writing the Meetings section, verify the exact categories returned for every candidate event. If WorkIQ does not return categories, run the category-only query above again. If the category still cannot be explicitly verified, exclude the event. Never substitute a general calendar-meeting or attendee-based result for this category filter.
- **WORKDAY BOUNDARY**: Only include meetings that start before 10:00 PM PDT on the specified date. Exclude meetings that start at or after 10:00 PM PDT.
- For each Required Meeting with other attendees:
  - Extract: Time (already in PDT), Meeting Name, Organizer details
  - If meeting notes are available in the response, include a brief 2-3 sentence summary; otherwise use "No notes recorded"
- Format: 
  - If Required Meetings exist: Table with columns [Time (PDT) | Meeting | Summary]
  - If NO Required Meetings found: Show text "No required meetings scheduled" (not a table)
- **IMPORTANT**: Do NOT include personal time blocks, focus blocks, lunch blocks, or self-created calendar holds—only actual meetings with other attendees

### 2. Completed Work
**Tool**: GitHub CLI graphql queries - devrel repo (https://github.com/github/devrel/issues)
- Query: Issues closed by user "aprilgittens" on the specified date
- **Important**: GitHub stores times in UTC. To capture your PDT workday (through 10PM PDT), convert the date range:
  - Start: `YYYY-MM-DD` at 12:00 AM PDT = `YYYY-MM-DD 07:00 UTC`
  - End: `YYYY-MM-DD` at 10:00 PM PDT = `(YYYY-MM-DD+1) 05:00 UTC`
  - Example for July 8: Query `closed:2026-07-08T07:00Z..2026-07-09T05:00Z`
- Use PowerShell with GitHub CLI (set $env:GITHUB_TOKEN=$null to use keyring auth):
  ```powershell
  gh api graphql -f query='query { search(query:"repo:github/devrel author:aprilgittens state:closed closed:YYYY-MM-DDTHH:MMZ..YYYY-MM-DDTHH:MMZ", type:ISSUE, first:10) { edges { node { ... on Issue { title url createdAt closedAt } } } } }'
  ```
- Extract for each issue:
  - Title, Link (url), Created date (createdAt - convert to PDT format YYYY-MM-DD), Updated/Closed date (closedAt - convert to PDT format YYYY-MM-DD)
- Format: Table with columns [Title | Link | Created | Updated]
- Date format in table: YYYY-MM-DD (converted to PDT from UTC timestamps)
- If no completed issues, show "No issues closed today"

### 3. Work in Progress
**Tool**: GitHub CLI graphql queries - devrel repo (https://github.com/github/devrel/issues)
- Query: Open issues in devrel repo where aprilgittens is assignee
- Use PowerShell with GitHub CLI:
  ```powershell
  gh api graphql -f query='query { search(query:"repo:github/devrel assignee:aprilgittens state:open", type:ISSUE, first:10) { edges { node { ... on Issue { title url createdAt } } } } }'
  ```
- Extract for each issue:
  - Title, Link (url), Created date (createdAt)
- Format: Table with columns [Title | Link | Created]
- Date format: YYYY-MM-DD
- Note: No date filtering - include all open issues regardless of creation date
- If no open issues assigned, show "No issues in progress"

### 4. Events
**Tool**: GitHub CLI graphql queries - IRL repo (https://github.com/github/IRL/issues)
- Query: Open issues in IRL repo where aprilgittens is assignee
- Use PowerShell with GitHub CLI:
  ```powershell
  gh api graphql -f query='query { search(query:"repo:github/IRL assignee:aprilgittens state:open", type:ISSUE, first:10) { edges { node { ... on Issue { title url } } } } }'
  ```
- Extract for each issue:
  - Issue name (title), Link (url)
- Format: Bullet list with format: **[Issue Name](link)**
- If no open issues assigned, show "No events assigned"

### 5. Community Insights
**Tool**: WorkIQ `workiq-ask` - OneNote
- Use `workiq-ask` to query OneNote for the week containing the specified date
- **Ask**: "Search ONLY my OneNote notebooks and pages for Community Insights recorded during the week containing [DATE]. Do not use Viva Engage, Teams, email, files, calendar, or organization-wide search results. Return each matching OneNote page title, its notebook/section, the recorded topic, observation, and source. If no non-template entries exist, say exactly: No Community Insights recorded for this week."
- Pass `timeZone: "America/Los_Angeles"` for consistency
- **OneNote verification gate**: Include an insight only when the WorkIQ response explicitly identifies its OneNote page title and notebook or section. Reject entries from Viva Engage, Teams, email, SharePoint files, calendar, or any other source, even if their text mentions "Community Insights." If every result is rejected or no non-template entry is returned, write "No Community Insights recorded for this week."
- For each entry returned, extract:
  - Topic (entry title/topic)
  - Observation (summary of the insight/observation)
  - Source (where the insight came from - link, person, document, etc.)
- Format: Bullet list with format:
  ```
  - **Topic**
    - Observation: [summary]
    - Source: [source reference]
  ```
- If no Community Insights recorded for that week or if only template placeholders exist, show "No Community Insights recorded for this week"

### 6. Overview
**Tool**: Auto-generated based on Completed Work and Work in Progress sections
- Synthesize the data from Completed Work and Work in Progress
- Create a 2-3 sentence narrative summary that:
  - Highlights key accomplishments from Completed Work
  - Describes current focus areas from Work in Progress
  - Reads naturally and flows as a cohesive summary
- Place at top of brief under the Date/Day headers
- Example style: "Today was focused on improving model selection workflows and automation. Closed issues on research workflow and LOE automation, while continuing work on prompt caching blog post and workshop updates. Current efforts are centered on content creation and internal tool improvements."

## Output Format

### File Location
- Create directory structure: `Briefs/Daily/{year}/{month}/`
  - Example: `Briefs/Daily/2026/06/`
- Filename: `daily-brief-{month}-{day}.md`
  - Example: `daily-brief-06-26.md`

### Template Structure
Follow the format exactly:
1. **Header**: Title with date (YYYY-MM-DD) and Day (e.g., "# Daily Brief — Friday, June 26, 2026")
2. **Overview**: Auto-generated 2-3 sentence summary
3. **Meetings**: Table or empty message
4. **Completed Work**: Table or empty message
5. **Work in Progress**: Table or empty message
6. **Events**: Bullet list or empty message
7. **Community Insights**: Bullet list or empty message
8. **Footer**: Generated timestamp in format `- Generated: YYYY-MM-DD HH:mm PDT`

### Timestamp Handling
- Convert all times to PDT (UTC-7)
- Use format: `YYYY-MM-DD HH:mm PDT` for generation timestamp
- Meeting times should display with PDT indicator (e.g., "4:00 PM – 4:30 PM PDT")
- Apply timezone conversion to all meeting times and event timestamps

## Error Handling
- **Missing Data**: If a section has no data, show "No items" or leave section empty (use template as guide)
- **Invalid Date**: If user provides invalid date, ask for clarification
- **API Failures**: If a data source is unavailable, note it in the section (e.g., "Unable to retrieve calendar data")
- **Missing OneNote/Calendar**: Handle gracefully with clear message about missing data
- **GitHub API Errors**: Provide user-friendly error message

## Credentials/Usernames
- GitHub user for filtering: **aprilgittens**
- WorkIQ will use stored Microsoft To Do credentials (SessionID: redacted, UserID: redacted)
- Timezone preference: **PDT (UTC-7)**

## Execution Checklist
Before generating the brief, verify:
- [ ] **READ THESE INSTRUCTIONS** - This entire file has been read and understood
- [ ] Date is confirmed/parsed correctly (or ask user for clarification)
- [ ] GitHub auth verified: Run `gh auth status` - should show aprilgittens authenticated via keyring
- [ ] All five data sources are queried:
  - [ ] Meetings (WorkIQ calendar + OneNote)
  - [ ] Completed Work (GitHub devrel repo closed issues)
  - [ ] Work in Progress (GitHub devrel repo open issues)
  - [ ] Events (GitHub IRL repo open issues)
  - [ ] Community Insights (OneNote weekly page)
- [ ] Timezone conversions verified: All times displayed in PDT with indicator
- [ ] Every listed meeting has an explicitly returned category containing the exact `Required Meeting` label; events with missing or unverified categories are excluded
- [ ] Every listed Community Insight explicitly identifies a OneNote page title and notebook/section; broad WorkIQ matches and non-OneNote sources are excluded
- [ ] Data is formatted according to template specifications
- [ ] Overview summary is generated (2-3 sentences synthesizing Completed + In Progress work)
- [ ] Output directory structure exists: `Briefs/Daily/{year}/{month}/`
- [ ] File is saved with correct naming: `daily-brief-{month}-{day}.md`
- [ ] Timestamp is in PDT: `YYYY-MM-DD HH:mm PDT`
- [ ] Confirmation message shows file path and summary of sections populated

## Example Response Flow
1. User: "Generate my daily brief"
2. **READ these instructions completely**
3. Confirm date with user if unspecified (or parse relative dates like "yesterday")
4. Verify GitHub authentication with `gh auth status`
5. Query all five data sources in parallel (when possible):
   - WorkIQ (via `workiq-ask`): Calendar meetings with timeZone="America/Los_Angeles", requiring explicit `Required Meeting` category verification, and OneNote community insights
   - GitHub CLI: devrel closed issues, devrel open issues, IRL open issues
6. Generate Overview based on completed and in-progress work (2-3 sentences)
7. Verify all times are displayed in PDT (workiq-ask handles this with timeZone parameter)
8. Compile all sections into markdown following template exactly
9. Create Briefs directory structure if needed
10. Save file to Briefs/Daily/{year}/{month}/daily-brief-{month}-{day}.md
11. Confirm to user: "✅ Daily brief generated and saved to Briefs/Daily/{year}/{month}/daily-brief-MM-DD.md" with section summary

## Troubleshooting
- **GitHub auth fails**: Run `$env:GITHUB_TOKEN = $null` before queries to use keyring authentication
- **WorkIQ direct fetch fails**: Use `workiq-ask` instead of direct `workiq-fetch` or `workiq-call_function` paths - direct WorkIQ fetch paths may have access restrictions. Always use `workiq-ask` for natural language queries to calendar and OneNote.
- **Timezone confusion**: 
  - Always pass `timeZone: "America/Los_Angeles"` to `workiq-ask` - it automatically converts times to PDT
  - For GitHub queries, remember to convert PDT date boundaries to UTC (PDT + 7 hours = UTC). Workday ends at 10PM PDT = 5AM UTC next day
  - All displayed times should be in PDT with indicator (e.g., "8:00 PM PDT")
- **Required Meeting category missing**: Do not infer that an event is required from its attendees or title. Re-run the category-only calendar query requesting exact categories; if the label cannot be explicitly confirmed, exclude the event and use "No required meetings scheduled" when no verified events remain.
- **Missing data sections**: Show "No [section] found" rather than empty tables for clarity

---
name: weekly-brief
description: Generate a comprehensive weekly summary synthesizing daily briefs into one cohesive overview
---

# Weekly Brief Skill

## Purpose
Aggregate daily briefs from a specified date range into a comprehensive weekly summary that synthesizes the week's work with an LLM overview and consolidated sections.

## Trigger
- "Weekly brief"
- "Generate weekly brief"
- "Create weekly summary"
- "Weekly debrief"
- And similar variations

## Execution Workflow

### 1. Prompt for Date Range
- Ask user: "What date range would you like to summarize? (format: MM-DD - MM-DD)"
- Example: "06-29 - 07-02"
- Parse the input to extract start date and end date
- Determine year: Use current year for both dates
- Handle cross-month ranges automatically (e.g., 06-29 - 07-02 spans June and July)

### 2. Date Range Parsing Logic
- Input format: MM-DD - MM-DD
- Extract: start_month, start_day, end_month, end_day
- Year = current year (from current_datetime)
- Validate dates exist in the identified months
- Create list of all dates between start and end (inclusive)
- Determine unique month/year combinations needed to search daily briefs

### 3. Daily Brief File Discovery
**Tool**: File system search
- For each month/year combination identified in the date range:
  - Search: `Briefs/Daily/{year}/{month}/daily-brief-{MM-DD}.md`
  - Example: For date range 06-29 - 07-02:
    - Search `Briefs/Daily/2026/06/` for daily-brief-06-29.md and daily-brief-06-30.md
    - Search `Briefs/Daily/2026/07/` for daily-brief-07-01.md and daily-brief-07-02.md
- Collect all matching files in chronological order
- If no files found, notify user and stop

### 4. Parse Daily Brief Files
For each daily brief file found (in chronological order):
- Extract the following sections:
  - **Overview**: Full text of overview paragraph
  - **Meetings**: Parse table, extract all rows (Time, Meeting, Summary)
  - **Completed Work**: Parse table, extract all rows (Title, Link, Created, Updated)
  - **Work in Progress**: Parse table, extract all rows (Title, Link, Created)
  - **Events**: Parse bullet list, extract all items
  - **Community Insights**: Parse bullet list, extract all items (Topic, Observation, Source)

### 5. Aggregate & Deduplicate Sections

#### Meetings (Chronological, No Dedup)
- Combine all meetings from all daily briefs
- Sort by time in chronological order (earliest first)
- Preserve all meetings even if repeated across days
- Format: Table with columns [Time (PDT) | Meeting | Summary]

#### Completed Work (Deduplicate by Title+Link)
- Combine all completed work items from all daily briefs
- Deduplicate: If same title AND link appear multiple times, keep only one
- Preserve order: Keep first occurrence
- Format: Table with columns [Title | Link | Created | Updated]

#### Work in Progress (Deduplicate by Title+Link)
- Combine all WIP items from all daily briefs
- Deduplicate: If same title AND link appear multiple times, keep only one
- Preserve order: Keep first occurrence
- Format: Table with columns [Title | Link | Created]

#### Events (Deduplicate)
- Combine all events from all daily briefs
- Deduplicate: Remove duplicate entries (match by title or description)
- Preserve order: Keep first occurrence
- Format: Bullet list

#### Community Insights (Deduplicate by Topic)
- Combine all community insights from all daily briefs
- Deduplicate: If same topic appears multiple times, keep all unique observations grouped under topic
- Format: Bullet list with format:
  ```
  - **Topic**
    - Observation: [summary]
    - Source: [source reference]
  ```

### 6. Generate Overview via LLM
**Tool**: Anthropic Claude API (via Copilot context)
- Input: All daily overview paragraphs from the date range
- Prompt: "Synthesize these daily work overviews into a single 2-3 sentence weekly summary that captures key themes, accomplishments, and current focus areas for the week of {start_date} to {end_date}. Be concise and highlight strategic activities."
- Output: 2-3 sentence weekly overview
- Fallback: If LLM call fails, concatenate daily overviews with " | " separator

### 7. Build Weekly Brief Output

Structure (in order):
1. **Header**: "# Weekly Brief — Week of {Start Date} to {End Date}"
   - Format: "# Weekly Brief — Week of Monday, June 29 to Wednesday, July 2, 2026"
   
2. **Needs from Manager** (NEW)
   - Content: Placeholder table
   - Format:
     ```
     | Request | Due Date |
     |---------|----------|
     |         |          |
     ```

3. **Overview**
   - Content: LLM-generated 2-3 sentence summary

4. **Meetings**
   - Content: All meetings in chronological order
   - Format: Table [Time (PDT) | Meeting | Summary]
   - If no meetings: "No meetings recorded"

5. **Completed Work**
   - Content: Deduplicated completed work items
   - Format: Table [Title | Link | Created | Updated]
   - If no items: "No issues closed during this period"

6. **Work in Progress**
   - Content: Deduplicated WIP items
   - Format: Table [Title | Link | Created]
   - If no items: "No issues in progress"

7. **Events**
   - Content: Deduplicated events
   - Format: Bullet list
   - If no items: "No events scheduled"

8. **Community Insights**
   - Content: Deduplicated insights
   - Format: Bullet list (Topic, Observation, Source)
   - If no items: "No Community Insights recorded"

9. **Interesting Reads** (NEW - Placeholder)
   - Content: Placeholder table
   - Format:
     ```
     | Article | Insight |
     |---------|---------|
     |         |         |
     ```

10. **Footer**
    - Content: Generated timestamp
    - Format: `- Generated: YYYY-MM-DD HH:mm PDT`

### 8. Write Output File

**File Location**:
- Directory: `briefs/weekly/{current-year}/{end-month}/`
  - Example: `briefs/weekly/2026/07/` (current year is 2026, end date month is 07)
- Filename: `weekly-brief-{start-date}-to-{end-date}.md`
  - Example: `weekly-brief-06-29-to-07-02.md`

**Actions**:
- Create markdown file with the weekly brief content
- Open file in editor for user review/editing
- Confirm: "✅ Weekly brief generated: briefs/weekly/{year}/{month}/weekly-brief-{start}-to-{end}.md"

## Output Format Example

```markdown
# Weekly Brief — Week of Sunday, June 29 to Wednesday, July 2, 2026

## Needs from Manager

| Request | Due Date |
|---------|----------|
|         |          |

## Overview

This week focused on developer relations and strategic initiatives around GitHub Copilot and FinOps. Key activities included conference attendance, thought leadership on AI cost management, and advocacy team alignment. Work in progress spans speaking engagements, content development on budget management and out-of-budget experiences, and ongoing contributions to the UBB/TBB content epic.

## Meetings

| Time (PDT) | Meeting | Summary |
|-----------|---------|---------|
| 12:00 PM – 1:00 PM | [Research report] FinOps X 2026: the state of FinOps & AI | Discussed FinOps X conference recap and industry trends around AI cost management. Covered competitive analysis and recommendations for Azure. |
| 3:00 PM – 3:45 PM | Advocacy Team Meeting | Strategic alignment on key initiatives and messaging. |

## Completed Work

No issues closed during this period.

## Work in Progress

| Title | Link | Created |
|-------|------|---------|
| Speaking: GitHub Copilot Day - Introduction to GitHub Copilot CLI | [#5467](https://github.com/github/devrel/issues/5467) | 2026-06-10 |
| Content: What happens when budgets are exhausted? | [#5282](https://github.com/github/devrel/issues/5282) | 2026-05-06 |

## Events

- **[Aug 12-13, 2026 : RenderATL 2026](https://github.com/github/IRL/issues/1926)**

## Community Insights

- **Companies cap AI usage**
  - Observation: Tesla has capped their employee AI spend at $200/week. This comes after the company's adoption push.
  - Source: The Information

## Interesting Reads

| Article | Insight |
|---------|---------|
|         |         |

---

- Generated: 2026-07-02 10:49 PDT
```

## Error Handling
- **Invalid Date Format**: If user provides date in wrong format, ask for clarification with example
- **No Files Found**: If no daily briefs exist for the specified date range, notify user and suggest checking file locations
- **Incomplete Date Range**: If some dates have no corresponding daily brief files, note which dates are missing and proceed with available data
- **LLM Synthesis Failure**: If Claude API fails, fall back to concatenating daily overviews
- **File Write Error**: If output file cannot be written, display error and ask for alternative location

## Timezone Handling
- All times from daily briefs are already in PDT (from daily brief generation)
- Preserve all times as-is when aggregating
- Use current PDT time for generation timestamp
- Timezone indicator (PDT) should be displayed in all time fields

## Execution Checklist
Before generating the weekly brief, verify:
- [ ] Date range is parsed correctly (MM-DD - MM-DD format)
- [ ] Start and end dates are identified
- [ ] Current year is correctly determined
- [ ] All daily brief files for the date range are discovered
- [ ] All sections are extracted and parsed correctly
- [ ] Deduplication logic is applied to all sections
- [ ] LLM overview synthesis is successful (or fallback used)
- [ ] Output file path is correct: `briefs/weekly/{current-year}/{end-month}/weekly-brief-{start}-to-{end}.md`
- [ ] All sections are formatted according to template
- [ ] Timestamp is in PDT format
- [ ] File is created and opened for user

## Notes on Implementation
- This is a data aggregation + light synthesis skill
- LLM is used ONLY for Overview synthesis; all other sections are aggregated as-is
- Deduplication ensures the weekly brief is concise and avoids redundancy
- The "Needs from Manager" table is a new section specific to weekly briefs
- "Interesting Reads" becomes a placeholder table instead of content aggregation
- Maintain chronological order for meetings (important for context)
- All other aggregated sections should be deduplicated to improve clarity