Weekly Summary Workflow

This repository stores the version 1 assets for generating weekly work summaries from:

• GitHub activity exports
• Local exported AI conversations
• Optional employee correction notes

Files

• docs/plans/2026-03-10-weekly-summary-design.md
  • Approved product and prompt design
• docs/plans/2026-03-10-weekly-summary-implementation.md
  • Task-by-task implementation plan
• docs/templates/weekly-summary-report.md
  • Human-readable Markdown weekly summary template
• docs/templates/weekly-summary-detail.csv
  • Structured detail table template
• prompts/weekly-summary-system-prompt.md
  • Main AI instruction block
• prompts/weekly-summary-input-template.md
  • Standard raw input wrapper
• examples/weekly-summary/github-activity-example.md
  • Example GitHub source records
• examples/weekly-summary/conversation-example.md
  • Example AI conversation source records
• scripts/README.md
  • Reserved entry points for future automation
• scripts/build-weekly-summary-input.ps1
  • Build one normalized raw input file from GitHub, conversation, and manual note exports
• scripts/build-weekly-summary-package.ps1
  • Build one complete AI-ready package that includes the system prompt, raw input, and output templates
• scripts/normalize-github-export.ps1
  • Convert raw GitHub CSV exports into the standard [GitHub] record format
• scripts/normalize-conversation-export.ps1
  • Convert raw conversation CSV exports into the standard [Conversation] record format
• scripts/run-weekly-summary-pipeline.ps1
  • Run the full local pipeline from raw CSV exports to the final AI-ready package
• scripts/collect-private-github-local.ps1
  • Collect private local git commits into the raw GitHub CSV schema
• scripts/merge-github-activity.ps1
  • Merge local git activity with GitHub API activity while deduplicating commits
• scripts/run-weekly-summary-private-pipeline.ps1
  • Run the private-enhanced local pipeline from local git repos to the final AI-ready package
• scripts/collect-private-github-api.ps1
  • Collect private GitHub pull requests and pull request reviews by using GITHUB_TOKEN
• scripts/generate-weekly-summary-report.ps1
  • Automatically generate final Markdown weekly summary and structured detail CSV from merged activity

Recommended Workflow

1. Collect the employee's weekly GitHub activity and normalize it to the example record format.
2. Export the employee's AI conversation log to plain text or Markdown and normalize it to the example record format.
3. Add optional employee notes for missing context, corrected completion times, and work not visible in GitHub.
4. Paste the standardized input into prompts/weekly-summary-input-template.md.
5. Use the full text from prompts/weekly-summary-system-prompt.md as the AI system prompt.
6. Ask the AI to produce:
   • A Markdown weekly summary
   • A structured detail table
   • A meeting discussion outline
7. Compare the AI output against the two templates in docs/templates.
8. Ask the employee to confirm any item marked as pending employee confirmation.

Time Confirmation Rules

Each work item should preserve two time fields:

• system_event_time
  • GitHub event time or conversation timestamp
• actual_completion_time
  • Employee-confirmed completion time when available

Use time_confidence to record certainty:

• high
• medium
• low

Do not invent completion times. Leave the field blank and mark low confidence when evidence is missing.

Output Expectations

The final weekly summary should:

• Group records by work item, not by commit
• Surface high-priority items
• Surface discussion-needed items
• Surface resource escalation needs
• Be readable by a direct manager without opening raw source logs

The structured detail table should:

• Keep one row per work item
• Preserve source traceability through github_refs and conversation_refs
• Match the field names used by the prompt contract

Version 1 Boundaries

Version 1 is intentionally document-first:

• No Feishu integration
• No required GitHub API access
• No required scheduled automation

This keeps the prompt and schema stable before building scripts.

Future Automation Notes

When scripts are added later, they must preserve:

• The raw input schema from prompts/weekly-summary-input-template.md
• The field names in docs/templates/weekly-summary-detail.csv
• The output sections defined by prompts/weekly-summary-system-prompt.md

Automation should extend this workflow, not redefine it.

Local Script

Use the script below to generate one raw input file that matches the prompt contract:

powershell -ExecutionPolicy Bypass -File .\scripts\normalize-github-export.ps1 `
  -InputFile .\tests\fixtures\raw-github-export.csv `
  -OutputFile .\tests\output\normalized-github-export.md

powershell -ExecutionPolicy Bypass -File .\scripts\normalize-conversation-export.ps1 `
  -InputFile .\tests\fixtures\raw-conversation-export.csv `
  -OutputFile .\tests\output\normalized-conversation-export.md

powershell -ExecutionPolicy Bypass -File .\scripts\build-weekly-summary-input.ps1 `
  -EmployeeName "Alice Zhang" `
  -StartDate "2026-03-09" `
  -EndDate "2026-03-15" `
  -GitHubFile .\tests\output\normalized-github-export.md `
  -ConversationFile .\tests\output\normalized-conversation-export.md `
  -ManualNotesFile .\tests\fixtures\manual-notes-example.md `
  -OutputFile .\tests\output\generated-weekly-summary-input.md

Use the script below to generate one complete AI package after the raw input file exists:

powershell -ExecutionPolicy Bypass -File .\scripts\build-weekly-summary-package.ps1 `
  -RawInputFile .\tests\output\generated-weekly-summary-input.md `
  -OutputFile .\tests\output\generated-weekly-summary-package.md

Or run the whole local pipeline in one command:

powershell -ExecutionPolicy Bypass -File .\scripts\run-weekly-summary-pipeline.ps1 `
  -EmployeeName "Alice Zhang" `
  -StartDate "2026-03-09" `
  -EndDate "2026-03-15" `
  -GitHubCsvFile .\tests\fixtures\raw-github-export.csv `
  -ConversationCsvFile .\tests\fixtures\raw-conversation-export.csv `
  -ManualNotesFile .\tests\fixtures\manual-notes-example.md `
  -OutputDirectory .\tests\output\pipeline

If you want to include private local repositories, use the private-enhanced pipeline:

powershell -ExecutionPolicy Bypass -File .\scripts\run-weekly-summary-private-pipeline.ps1 `
  -EmployeeName "Guanming Zhu" `
  -StartDate "2026-03-03" `
  -EndDate "2026-03-10" `
  -RepoPaths .\tests\output\private-pipeline-repo `
  -AuthorNames "Guanming Zhu" `
  -AuthorEmails "sam@example.com" `
  -GitHubLogin "SSSSamZhu" `
  -ConversationCsvFile .\tests\fixtures\empty-conversation-export.csv `
  -ManualNotesFile .\tests\fixtures\manual-notes-example.md `
  -OutputDirectory .\tests\output\private-pipeline

This private-enhanced path currently:

• Collects local private git commits from the repo paths you provide
• Can also collect private GitHub PR and review activity when GITHUB_TOKEN is available
• Merges local and API activity into the existing GitHub CSV shape
• Reuses the existing normalization and AI-package scripts
• Automatically generates final Markdown and detail outputs

If you want private PR and review activity, set the token in the same shell first:

$env:GITHUB_TOKEN="your_read_only_token"

Then pass explicit repo list values such as SSSSamZhu/SomniaAPP to the private-enhanced pipeline.