Daily Work Workflow

The standard workflow for daily development with perfect session continuity.

Overview

Start session → Review context → Work → Save frequently → Full save before break → Repeat

Goal: Never lose context, always know where you are, resume instantly.

Morning: Start of Session

1. Review Context (2-3 minutes)

/review-context

What you see:

• 📊 Quick Reference (project overview)
• 📝 Last session summary
• 🔄 Current work in progress
• 🚧 Active blockers
• ✅ Recent accomplishments
• 🔍 Version check

Example output:

📊 Quick Reference
Project: TaskFlow - Async task management
Phase: MVP Development
Current Focus: Stripe payment integration
Last Session: Session 12 (2025-10-23)

📝 Last Session Summary (Session 12)
Implemented Stripe webhook handler for payment confirmation.
Added idempotency checking to prevent duplicates.
Tested with Stripe CLI successfully.

🔄 Work In Progress
- Order confirmation email (next task)
- End-to-end testing (after email)

🚧 Blockers
None currently

✅ Recent Accomplishments
- Webhook handler implementation
- Idempotency system
- Order status updates

🔍 System Version
Current: v3.2.2 (latest)

2. Orient Yourself (1 minute)

Read the output and ask:

• ✅ What did I accomplish last session?
• ✅ What am I working on now?
• ✅ Are there any blockers?
• ✅ What's the next task?

Result: You're oriented and ready to start working immediately.

3. Open Relevant Files (1 minute)

Based on "Work In Progress", open the files you'll need:

# From review-context output:
# "Next task: Order confirmation email"

# Open relevant files:
code app/api/webhooks/stripe/route.ts
code lib/email/orderConfirmation.ts
code context/DECISIONS.md  # Check DEC-014 about email choice

Working: Active Development

Save Frequently (Every 30-60 minutes)

/save

When to save:

• ✅ Finished a feature
• ✅ Fixed a bug
• ✅ About to take a break
• ✅ Before switching contexts
• ✅ After significant progress
• ✅ Just finished a good coding session

What /save does:

1. Updates STATUS.md with current work
2. Regenerates Quick Reference
3. Captures work in progress
4. Documents blockers if any
5. Takes 2-3 minutes

Example workflow:

# 9:00 AM - Start work
/review-context

# 9:05 AM - 10:30 AM
# Implement order confirmation email
# Write tests
# Debug send grid integration

# 10:30 AM - Save progress
/save

# AI updates STATUS.md:
## Work In Progress
- ✅ Order confirmation email template created
- ✅ SendGrid integration configured
- 🔄 Email trigger from webhook (in progress)
- ⏳ Test email delivery

Handle Blockers

When you encounter a blocker:

1. Document it immediately:

/save

2. AI prompts for blocker details:

Blocker detected. Describe the issue:
> Webhook fires but email doesn't send. No errors in logs.

Impact:
> Can't test end-to-end flow without email confirmation

Options being considered:
> 1. Add more detailed logging
> 2. Test SendGrid directly (bypass webhook)
> 3. Check SendGrid dashboard for failures

3. STATUS.md updated:

## Blockers & Decisions

**BLOCKER-002: Email Not Sending from Webhook**
- Issue: Webhook processes successfully but email not sent
- Impact: Can't test full checkout flow
- Investigation: No errors in logs, need better visibility
- Options: Add logging, test SendGrid separately, check dashboard
- Blocking: End-to-end testing

4. Work on blocker or switch tasks:

# Either:
# - Investigate blocker (add logging, test separately)
# - Switch to different task (end-to-end testing can wait)

# Document what you're doing:
/save

Make Decisions

When you make a technical decision:

1. Add to DECISIONS.md:

## DEC-016: Use SendGrid API v3

**Decision:** Use SendGrid API v3 (not v2)

**Date:** 2025-10-24

**Rationale:**
- v3 is current version (v2 deprecated 2024)
- Better TypeScript support
- Cleaner API (JSON instead of form-encoded)

2. Reference in STATUS.md:

/save

# STATUS.md shows:
### Decisions Made This Session
- DEC-016: SendGrid API v3 (see DECISIONS.md)

Midday: Before Break

Full Save Before Lunch (10-15 minutes)

/save-full

Why full save:

• Comprehensive SESSIONS.md entry
• Mental models documented
• Problem-solving approach captured
• Git operations auto-logged
• Decision rationale preserved

What happens:

1. Everything /save does (STATUS.md update)
2. Creates new SESSIONS.md entry
3. AI documents:
   - Mental models (current understanding)
   - What changed (features, bugs, refactoring)
   - Decisions made (with links)
   - Problem-solving approach
   - Files changed
   - Git operations (auto-logged)
   - Next session start point
4. Takes 10-15 minutes

Example SESSIONS.md entry:

## Session 13 | 2025-10-24 | Order Confirmation Email

**Duration:** 3h | **Focus:** Email notifications | **Status:** ✅ Complete

### TL;DR
Implemented order confirmation email triggered from Stripe webhook.
Integrated SendGrid API v3, created email template, tested delivery.
Email sends successfully after payment confirmation.

### Mental Models

**Current understanding:**
- Webhook confirms payment → triggers email send
- SendGrid API v3 for email delivery
- Template includes order details, receipt, tracking
- Error handling: log failures, retry 3x with backoff

[... rest of entry ...]

### Next Session
Start with end-to-end testing:
1. Full checkout flow (cart → payment → email)
2. Verify email delivery
3. Test error cases

Take Break

• Lunch
• Meeting
• Context switch to different project
• End of day

You're safe to step away - full context preserved in SESSIONS.md.

Afternoon: Resume After Break

Review Context Again (2 minutes)

/review-context

What you see:

• Last session: Session 13 (morning work)
• Current work: End-to-end testing
• Next steps: Full checkout flow test

Result: Pick up exactly where you left off, zero context loss.

Continue Working

Same pattern as morning:

• Work on tasks
• /save frequently (30-60 min)
• Document blockers
• Make decisions
• Test features
• Write code

End of Day: Comprehensive Save

Full Save Before Ending (10-15 minutes)

/save-full

Creates Session 14 entry:

## Session 14 | 2025-10-24 | End-to-End Testing & Bug Fixes

**Duration:** 2.5h | **Focus:** Testing and debugging | **Status:** ✅ Complete

### TL;DR
Completed end-to-end testing of checkout flow. Found and fixed
email template rendering bug. Verified full flow works: cart →
checkout → payment → email. Ready for staging deployment.

[... mental models, changes, etc ...]

### Next Session
Deploy to staging:
1. Push to main branch
2. Verify Vercel deployment
3. Test on staging.taskflow.app
4. Check Stripe test mode webhooks

Review Tomorrow's Start Point

# Read SESSIONS.md entry you just created
# Verify "Next Session" is clear
# Ensure you'll know where to start tomorrow

Push to GitHub (if ready)

git add .
git commit -m "Add order confirmation email + end-to-end tests"
git push origin main

# Git operations auto-logged in SESSIONS.md

Real-World Example: Full Day

9:00 AM - Start

/review-context

# Output:
Project: TaskFlow
Last Session: Session 12 (Webhook implementation)
Current Focus: Order confirmation email
Next Task: Trigger email from webhook

9:05 AM - 10:30 AM: Work

• Implement email trigger
• Write SendGrid integration
• Create email template
• Add error handling

10:30 AM: Save

/save

# STATUS.md updated:
## Work In Progress
- ✅ Email template created
- ✅ SendGrid integration done
- 🔄 Testing email delivery

10:30 AM - 12:00 PM: More Work

• Test email with real data
• Fix template rendering bug
• Verify email looks good

12:00 PM: Lunch - Full Save

/save-full

# Creates Session 13 entry
# Documents morning's work
# Captures mental models
# Auto-logs git commits

12:00 PM - 1:00 PM: Lunch Break

Walk away confident - full context saved.

1:00 PM: Resume

/review-context

# Output:
Last Session: Session 13 (Email implementation)
Current Focus: End-to-end testing
Next Task: Test full checkout flow

1:05 PM - 3:00 PM: Work

• Full checkout flow test
• Test error cases
• Fix bugs found
• Verify webhooks reliable

3:00 PM: Save

/save

# STATUS.md updated with test results

3:00 PM - 5:00 PM: More Work

• Write additional tests
• Refactor email service
• Update documentation
• Prep for staging deploy

5:00 PM: End of Day - Full Save

/save-full

# Creates Session 14 entry
# Documents afternoon's work
# Sets up tomorrow's start point

5:15 PM: Done

Leave confident - tomorrow you'll pick up instantly.

Tips for Daily Workflow

1. Start Every Session with /review-context

# ALWAYS first command:
/review-context

# Even if you just left 10 minutes ago
# Builds the habit
# Ensures you never miss context

2. Save More Than You Think You Need

# Don't just save at breaks
# Save after meaningful progress

# Good times to save:
- Finished implementing a feature
- Fixed an annoying bug
- Made a decision
- Before switching tasks
- Just completed good work

# It's fast (2-3 min) and safe

3. Use /save-full at All Boundaries

# Lunch breaks
/save-full

# Meetings
/save-full

# End of day
/save-full

# Context switches
/save-full

# Before big refactors
/save-full

4. Read Your SESSIONS.md Entries

# After /save-full, read what was created:
cat context/SESSIONS.md | tail -100

# Verify:
# - Mental models accurate?
# - Next session clear?
# - Problem-solving captured?

# Make manual edits if needed
# This is YOUR history

5. Trust the Process

# System captures more than you think:
- Git operations (auto-logged)
- File changes (extracted from commits)
- Decision rationale (from your input)
- Mental models (AI documents understanding)

# You don't need to remember everything
# The system remembers for you

Success Metric

"I can end work abruptly, start days later, run /review-context, and be productive within 5 minutes."

When this is true, daily workflow is working.

Next Steps

• /save Command - Quick saves
• /save-full Command - Comprehensive saves
• /review-context Command - Session start
• AI-to-AI Handoff - Project handoffs