--- name: morning-standup-brief-generator description: "Compile local notes, tasks, and project files into a structured standup brief — no connectors required. Use when preparing for morning standup, daily sync, or team check-in. Triggers: 'morning standup brief', 'daily brief', 'prepare me for standup', 'подготовь меня к стендапу', 'дейли бриф', 'утренний бриф'." version: 1.0.0 --- # Morning Standup Brief Generator This skill compiles local notes, task files, and project documents into a structured daily standup brief in the format: **Yesterday / Today / Blockers / Questions**. It works entirely from local files — no MCP connectors, external APIs, or technical setup required. **Input:** - Path to a working directory (or Cowork session folder) containing `.md` and/or `.txt` files **Output:** - Standup brief displayed in chat - File `standup-brief-YYYY-MM-DD.md` saved to the working directory --- ## Language Detection Detect the user's language from their message: - If Russian (or contains Cyrillic): respond in Russian - If English (or other Latin-script language): respond in English - If ambiguous: respond in the language of the trigger phrase used --- ## Instructions ### Step 1: Validate Input 1. Check if the user has provided a working directory path. - If no path is provided and no Cowork session folder is active: stop and report: "Please provide the path to your working folder. Example: /Users/yourname/work-notes/" - If path is provided but does not exist: report "Folder not found: [path]. Please check the path and try again." 2. Check for readable files (`.md` and `.txt`) in the directory. - If the folder exists but contains no `.md` or `.txt` files: report "No readable files found (.md, .txt). Add notes or task files to the folder and try again." 3. Check if a `standup-brief-YYYY-MM-DD.md` file for today's date already exists. - If it exists: ask "A standup brief for today already exists. Overwrite it?" - If user confirms: proceed. If user declines: stop cleanly. ### Step 2: Discover and Filter Files 1. List all `.md` and `.txt` files in the directory (non-recursive; top level only). 2. Apply time filter: prioritize files **modified in the last 48 hours**. - If no files modified in last 48 hours: include all files and note "No recent changes found — using all files in folder." - If more than 20 files match: limit to those modified in the last 48 hours and notify: "Found [N] files — filtered to [M] recently modified (last 48h)." 3. Build a list of files to process. For each file, note its name and modification date. ### Step 3: Extract Task Signals Read each selected file and extract the following: 1. **Completed items** (for Yesterday section): - Lines containing `- [x]`, `[x]`, `✓`, `DONE`, `COMPLETED` - Include the item text (trim markers) 2. **Open tasks** (for Today section): - Lines containing `- [ ]`, `[ ]`, `TODO`, `#task`, `#todo` - Order by earliest deadline if dates are present (e.g., `(due YYYY-MM-DD)`, `by [date]`) 3. **Blockers** (for Blockers section): - Lines containing `BLOCKER`, `BLOCKED`, `#blocked`, `waiting on`, `waiting for`, `depends on`, `stuck` - Also: open tasks with no progress markers that appear in multiple files or are repeated 4. **Questions / Awaiting response** (for Questions section): - Lines containing `?`, `ASK`, `#question`, `waiting for [name]`, `@[name]`, `ожидаю`, `вопрос:` **Edge Cases:** - If a file contains no structured task markers: extract bullet points and short sentences as candidate items; mark section as `(extracted from notes — no formal task markers found)`. - If a file contains mixed languages (EN + RU): process all markers in both languages; output in the dominant language of the file content. ### Step 4: Prioritize and Deduplicate 1. Remove duplicate items (same text appearing in multiple files — keep once). 2. Sort Today items: - Items with explicit deadlines first (nearest deadline at top) - Then remaining open tasks in order of file appearance 3. Limit sections to reasonable length: - Yesterday: max 7 items - Today: max 7 items - Blockers: list all (no cap) - Questions: list all (no cap) 4. If a section has no items: write `— none` under that section header. ### Step 5: Compose and Output Brief 1. Build the standup brief using the Output Format template below. 2. Insert today's date (YYYY-MM-DD) in the title. 3. Add a footer line listing all source files processed. 4. Display the full brief in chat. 5. Save as `standup-brief-YYYY-MM-DD.md` in the working directory. - Confirm save with: "Brief saved to: [path]/standup-brief-YYYY-MM-DD.md" --- ## Output Format ```markdown # Standup Brief — YYYY-MM-DD ## Yesterday (Done) - [completed item 1] - [completed item 2] ## Today (Plan) - [open task 1] - [open task 2] ## Blockers - [blocker 1] — none (if no blockers found) ## Questions / Awaiting Response - [question or pending action 1] — none (if nothing pending) --- *Generated: YYYY-MM-DD | Sources: [file1.md, file2.md, ...]* ``` **Field rules:** - Each item is one line, starting with `- ` - No sub-bullets (keep scannable for standup format) - Completed items: state what was finished, not the task description alone - Today items: actionable verb + object (e.g., "Review PR #42", "Send status update to Alex") - Blockers: name the blocker and who/what is blocking - Questions: include the recipient if known (e.g., "Awaiting answer from @Maria on API spec") --- ## Negative Cases - **No path provided + no active Cowork folder:** Stop. Report: "Please provide the path to your working folder." - **Folder not found at path:** Stop. Report: "Folder not found: [path]. Check the path and retry." - **Folder is empty (no .md/.txt):** Stop. Report: "No readable files (.md, .txt) found in [path]. Add notes or task files first." - **All files are binary or unreadable:** Skip unreadable files silently; if all files are unreadable, report: "Could not read any files in [path]. Files may be in unsupported formats." - **Today's brief already exists (user declines overwrite):** Stop cleanly. Report: "Keeping existing brief at [path]/standup-brief-YYYY-MM-DD.md."