Session Note

What I do

• Create .opencode/notes/ directory if it doesn't exist
• Generate timestamped markdown file: YYYY-MM-DD-HHmm-<description>.md
• Document files changed with diffs (truncated with git reference for full diff)
• Capture incomplete todos from TodoWrite if present
• Provide 2-3 sentence summary of where the session left off
• Multiple notes per session create new files with incremented timestamps

When to use me

Use this skill when:

• User explicitly requests a session note
• User wants to pause work and capture current state for later continuation
• User needs to document progress before switching tasks/projects

Note Creation Workflow

1. Create directory: mkdir -p .opencode/notes
2. Generate filename:
   • Extract main task from conversation context
   • Format: YYYY-MM-DD-HHmm-<brief-description>.md
   • Example: 2026-02-14-1630-add-csv-export.md
   • Description: kebab-case, max 4-5 words from main task
   • Fallback: work-session if task unclear
3. Gather context:
   • Scan conversation for accomplished work
   • Identify main goals and outcomes
4. Detect file changes:
   • If git repo: git status --short
   • Parse: M = modified, A/?? = created, D = deleted
   • If not git: extract from conversation (files agent wrote/edited)
5. Get diffs:
   • For modified files: git diff <file> (if git available)
   • For new files: read first 20-30 lines or key sections
   • Truncate large diffs (see diff handling below)
6. Extract todos:
   • Check if TodoWrite was used in session
   • Filter for status: pending or status: in_progress
   • Format as markdown checklist
7. Write note with structure below
8. Confirm: Tell user the note location

Filename Generation

• Pattern: YYYY-MM-DD-HHmm-<description>.md
• Timestamp: date +%Y-%m-%d-%H%M
• Description extraction:
  • Scan conversation for primary task/goal
  • Extract key verbs + objects: "add CSV export" → add-csv-export
  • Sanitize: lowercase, spaces→hyphens, remove special chars
  • Max 4-5 words, keep concise
• Examples:
  • 2026-02-14-1630-add-csv-export.md
  • 2026-02-14-1845-fix-memory-leak.md
  • 2026-02-14-2015-refactor-auth-module.md

Note Structure

# Session Note: <Descriptive Title>

**Date:** YYYY-MM-DD HH:mm
**Branch:** <current-branch> (only if git repo)

## Summary
<2-3 sentence overview of what was accomplished in this session>

## Files Changed

### Created
- `path/to/new-file1.ext`
- `path/to/new-file2.ext`

### Modified
- `path/to/modified-file1.ext`
- `path/to/modified-file2.ext`

### Deleted
- `path/to/deleted-file.ext`

(Only include sections that apply - skip empty categories)

## Changes Detail

### path/to/file1.ext
```diff
<diff output - first 50 lines>
... (diff truncated, run `git diff path/to/file1.ext` for full changes)

path/to/file2.ext

<diff output - first 50 lines>
... (diff truncated, run `git diff path/to/file2.ext` for full changes)

(Repeat for each modified file)

Tests/Validations

• npm test - All 47 tests passed
• npm run build - Build successful
• Manual testing: CSV export works with 1000+ rows

(Only include if tests/validations were run - omit section otherwise)

Incomplete Tasks

• Add error handling for edge cases
• Update documentation
• Review performance with larger datasets

(Only include if TodoWrite has incomplete items - omit section otherwise)

Where We Left Off

<2-3 sentences max describing current state, what was being worked on, and immediate context for continuation>

Next Steps

1. <actionable next step 1>
2. <actionable next step 2>
3. <actionable next step 3>

## Diff Handling

- **Size threshold**: If diff exceeds 50 lines:
  - Include first 50 lines only
  - Add note: `... (diff truncated, run \`git diff <file>\` for full changes)`
- **New files**: Include first 20-30 lines or key sections
- **Large new files** (>100 lines): Show structure/imports + key functions only
- **No git**: For files without git, describe changes textually:

path/to/file.ext

• Added new function processData()
• Updated error handling in validateInput()
• Refactored class structure

## File Change Detection

**With git:**
```bash
git status --short

Parse output:

• M file.ext → Modified
• A file.ext → Created (staged)
• ?? file.ext → Created (untracked)
• D file.ext → Deleted

Without git:

• Extract from conversation context
• Track files agent used Write/Edit tools on
• Note: "Files listed based on session activity (not a git repository)"

Todo Extraction

Process:

1. Check if TodoWrite tool was used in current session
2. Filter for incomplete items: status: "pending" or status: "in_progress"
3. Format as markdown checklist with - [ ] prefix
4. If no incomplete todos or TodoWrite not used → omit section

Example:

## Incomplete Tasks
- [ ] Add unit tests for CSV exporter
- [ ] Handle empty dataset edge case
- [ ] Update API documentation

Edge Cases

• No files changed: Note this in Summary, focus on planning/discussion done
• No git repo: Skip branch info, use conversation context for file tracking
• No todos: Omit "Incomplete Tasks" section entirely
• No tests run: Omit "Tests/Validations" section entirely
• Multiple calls in session: Create new note with current timestamp (naturally incremented)
• Very short session: Still create note, may have minimal content in some sections

Example Complete Note

# Session Note: Add CSV Export Functionality

**Date:** 2026-02-14 16:30
**Branch:** feat/csv-export

## Summary
Implemented CSV export functionality for user data with configurable column selection. Added validation for large datasets and basic error handling. Tests passing but documentation still needed.

## Files Changed

### Created
- `src/exporters/csv-exporter.ts`
- `tests/csv-exporter.test.ts`

### Modified
- `src/components/ExportButton.tsx`
- `src/types/export.types.ts`

## Changes Detail

### src/exporters/csv-exporter.ts
```diff
+export class CSVExporter {
+  constructor(private options: ExportOptions) {}
+
+  export(data: UserData[]): string {
+    const headers = this.getHeaders();
+    const rows = data.map(item => this.formatRow(item));
+    return [headers, ...rows].join('\n');
+  }
+
+  private getHeaders(): string {
+    return this.options.columns.join(',');
+  }
+
+  private formatRow(item: UserData): string {
+    return this.options.columns
+      .map(col => this.escapeValue(item[col]))
+      .join(',');
+  }
+}
... (diff truncated, run `git diff src/exporters/csv-exporter.ts` for full changes)

src/components/ExportButton.tsx

 export function ExportButton({ data }: Props) {
+  const handleExport = () => {
+    const exporter = new CSVExporter({ columns: ['name', 'email', 'role'] });
+    const csv = exporter.export(data);
+    downloadFile(csv, 'users.csv');
+  };
+
   return (
-    <button>Export</button>
+    <button as CSV</button>
   );
 }

Tests/Validations

• npm test - All 12 tests passed
• Manual testing: Exported 500 user records successfully

Incomplete Tasks

• Add error handling for invalid column names
• Update user documentation with export feature
• Add support for Excel format

Where We Left Off

CSV export is functional and tested. Error handling for edge cases (empty datasets, invalid columns) still needs implementation. Documentation update is next priority.

Next Steps

1. Implement error handling for edge cases
2. Add user documentation for export feature
3. Consider adding Excel export format support