/save and /resume Commands

Save current work context and resume previous sessions for seamless workflow continuity in Mega Brain

/save & /resume - Session Persistence

Save and restore work context between Claude Code sessions, preserving tasks, files, notes, and next steps.

Commands

/save

Save current session context

/resume

Restore previous session context

/save Command

Syntax

/save

No parameters required. Command is interactive.

What Gets Saved

Context Elements:
  - Last task description
  - Current status (in_progress | completed | blocked)
  - Pending items list
  - Next steps
  - Important file paths
  - User notes
  - Timestamp

Execution Flow

1. Read Current Session

   # Loads existing session file
   cat /logs/LIVE-SESSION/CURRENT-SESSION.md
   
2. Prompt User for Context

   What do you want me to save as context?
   
   Examples:
   - "Processing Alex Hormozi materials, 3 of 5 files done"
   - "Fixing DNA extraction bug, need to test with Cole Gordon"
   - "Setting up new expert: Sam Ovens"
   
3. Update Session Files

   Saves to two locations:

   CURRENT-SESSION.md (human-readable):

   # Session: 2026-03-06
   
   ## Current Task
   Processing Alex Hormozi materials through pipeline
   
   ## Status
   IN_PROGRESS
   
   ## Completed
   - ✓ Ingested AH024 (VSL on scaling)
   - ✓ Ingested AH025 (Podcast episode)
   - ✓ Processed AH024 through all 8 phases
   
   ## Pending
   - [ ] Process AH025
   - [ ] Process AH026
   - [ ] Extract DNA after 3rd file
   
   ## Next Steps
   1. /process-jarvis "inbox/ALEX HORMOZI/PODCASTS/AH025.txt"
   2. /process-jarvis "inbox/ALEX HORMOZI/MASTERCLASSES/AH026.txt"
   3. /extract-dna alex-hormozi
   
   ## Important Files
   - inbox/ALEX HORMOZI/VSL/AH024.txt
   - inbox/ALEX HORMOZI/PODCASTS/AH025.txt
   - inbox/ALEX HORMOZI/MASTERCLASSES/AH026.txt
   
   ## Notes
   AH024 had excellent heuristics on pricing. Make sure to review
   HEURISTICAS.yaml after DNA extraction.
   
   ---
   Saved: 2026-03-06T15:30:00Z
   

   CONTEXT.json (machine-readable):

   {
     "last_task": "Processing Alex Hormozi materials through pipeline",
     "status": "in_progress",
     "pending": [
       "Process AH025",
       "Process AH026",
       "Extract DNA after 3rd file"
     ],
     "next_steps": [
       "/process-jarvis \"inbox/ALEX HORMOZI/PODCASTS/AH025.txt\"",
       "/process-jarvis \"inbox/ALEX HORMOZI/MASTERCLASSES/AH026.txt\"",
       "/extract-dna alex-hormozi"
     ],
     "important_files": [
       "inbox/ALEX HORMOZI/VSL/AH024.txt",
       "inbox/ALEX HORMOZI/PODCASTS/AH025.txt",
       "inbox/ALEX HORMOZI/MASTERCLASSES/AH026.txt"
     ],
     "notes": "AH024 had excellent heuristics on pricing. Review HEURISTICAS.yaml after DNA extraction.",
     "timestamp": "2026-03-06T15:30:00Z"
   }
   
4. Confirm to User

   [SAVED] Context saved in:
   - /logs/LIVE-SESSION/CURRENT-SESSION.md
   - /logs/LIVE-SESSION/CONTEXT.json
   
   Next session: execute /resume to load context
   

Example Interaction

User: /save

Assistant: What do you want me to save as context?

User: I'm halfway through processing Cole Gordon materials. Processed
CG003 and CG004. Still need to do CG005, then extract DNA. CG004 had
great content on the Email Hammer framework.

Assistant: [SAVED] Context saved in:
- /logs/LIVE-SESSION/CURRENT-SESSION.md
- /logs/LIVE-SESSION/CONTEXT.json

Session Summary:
  • Task: Processing Cole Gordon materials
  • Status: In Progress
  • Pending: Process CG005, extract DNA
  • Note: Email Hammer framework in CG004

Next session: execute /resume to load context

/resume Command

Syntax

/resume

No parameters required.

What Gets Loaded

Restored Context:
  - Previous task description
  - Task status
  - Pending items
  - Next recommended steps
  - Important file references
  - User notes
  - Original save timestamp

Execution Flow

1. Load Session Files

   # Reads both files
   cat /logs/LIVE-SESSION/CURRENT-SESSION.md
   cat /logs/LIVE-SESSION/CONTEXT.json
   
2. Display Summary

   [RESUMING SESSION]
   
   ## Last Activity
   Processing Cole Gordon materials through pipeline
   
   ## Status
   In Progress
   
   ## Completed
   ✓ Processed CG003 (Masterclass - Scaling Sales Teams)
   ✓ Processed CG004 (Blueprint - Email Hammer Framework)
   
   ## Pending
   ◻ Process CG005
   ◻ Extract DNA (after 3+ sources)
   
   ## Next Steps
   1. /process-jarvis "inbox/COLE GORDON/COURSES/CG005.txt"
   2. /extract-dna cole-gordon
   
   ## Important Files
   • inbox/COLE GORDON/MASTERCLASSES/CG003.txt
   • inbox/COLE GORDON/BLUEPRINTS/CG004.txt
   • inbox/COLE GORDON/COURSES/CG005.txt
   
   ## Notes
   Email Hammer framework in CG004 has great heuristics.
   Review after DNA extraction.
   
   ---
   Last saved: 2026-03-06 15:30 (2 hours ago)
   
   Want to continue where you left off?
   
3. Await User Confirmation

   User can:

   • Proceed with suggested next steps
   • Do something else
   • Update the context

Example Interaction

User: /resume

Assistant:
[RESUMING SESSION]

## Last Activity
Processing Cole Gordon materials through pipeline

## Status
In Progress

## Completed
✓ Processed CG003 (Masterclass)
✓ Processed CG004 (Blueprint)

## Pending
◻ Process CG005
◻ Extract DNA

## Next Steps
1. /process-jarvis "inbox/COLE GORDON/COURSES/CG005.txt"
2. /extract-dna cole-gordon

## Notes
Email Hammer framework in CG004 - review after DNA extraction.

---
Last saved: 2 hours ago

Want to continue where you left off?

User: Yes, let's process CG005

Assistant: [Executes /process-jarvis command...]

Storage Location

logs/LIVE-SESSION/
├── CURRENT-SESSION.md     # Human-readable context
├── CONTEXT.json          # Machine-readable context
└── history/              # Previous sessions (archived)
    ├── 2026-03-05.md
    ├── 2026-03-04.md
    └── ...

Only the most recent session is kept in CURRENT-SESSION.md. Previous sessions are archived to history/ by date.

Session Status Values

in_progress

Meaning:
  Work is actively ongoing, not yet complete

Use when:
  - Processing multiple files (some done, some pending)
  - Multi-step task partially complete
  - Waiting for something (but will continue)

Example:
  "Processing 5 videos, 3 done, 2 remaining"

completed

Meaning:
  Task fully finished, no pending work

Use when:
  - All files processed
  - DNA extraction complete
  - Session concluded successfully

Example:
  "Extracted DNA for Cole Gordon, all 5 sources processed"

blocked

Meaning:
  Cannot proceed without external input/action

Use when:
  - Waiting for API keys
  - Missing source files
  - Dependency not met
  - Bug preventing progress

Example:
  "Pipeline stuck at Phase 4, duplicate detection failing"

Best Practices

1. Save at Natural Break Points

Good times to save:

End of work session:
  "/save" before closing Claude Code

Before switching tasks:
  Finish processing, then /save before starting new work

After completing a batch:
  Processed 3 files? Save before ingesting more.

2. Be Specific in Notes

Good vs bad notes:

Good

notes: |
  CG004 has Email Hammer framework (12 emails, 48h).
  Heuristic: "If show rate < 60%, deploy Email Hammer".
  This should map to LNS agent memory.

3. List Important Files

Include files you’ll need to reference:

important_files:
  - "inbox/EXPERT/file-with-good-heuristics.txt"
  - "knowledge/dna/expert/HEURISTICAS.yaml"  # For review
  - "processing/insights/INSIGHTS-STATE.json"  # Check contradictions

4. Clear Next Steps

Make it easy to resume:

Good:
  next_steps:
    - "/process-jarvis \"inbox/PERSON/file.txt\""
    - "/extract-dna person-name"
    - "Review HEURISTICAS.yaml for person-name"

Bad:
  next_steps:
    - "Finish processing"
    - "Do the DNA thing"

5. Update Status Regularly

If task status changes:

# Was: in_progress
# Now: blocked (found bug)
/save
# Update with new status and blocker details

Use Cases

Multi-Session Projects

Scenario:
  Processing 10 videos for new expert over 3 days

Day 1:
  - Ingest 10 videos
  - Process 3 videos
  - /save (status: in_progress, 7 pending)

Day 2:
  - /resume
  - Process 4 more videos
  - /save (status: in_progress, 3 pending)

Day 3:
  - /resume
  - Process final 3 videos
  - Extract DNA
  - /save (status: completed)

Blocked Workflows

Scenario:
  Pipeline fails mid-processing

Session 1:
  - Start processing CG005
  - Phase 4 fails (missing chunk_ids)
  - /save (status: blocked, note: "Phase 4 bug, needs fix")

Later (after fix):
  - /resume
  - See note about Phase 4 bug
  - Retry processing
  - /save (status: completed)

Handoff Between Sessions

Scenario:
  Different Claude Code sessions (new chat)

Old session:
  - Work for 2 hours
  - /save with detailed context
  - Close Claude Code

New session:
  - Open Claude Code (fresh chat)
  - /resume
  - Continue seamlessly

Troubleshooting

”No session to resume”

Issue: /resume finds no saved context

Solution:

# Check if files exist
ls -la logs/LIVE-SESSION/

# If missing, start fresh and save:
/save

“Session is stale”

Issue: Last save was weeks ago, context outdated

Solution:

# Archive old session
mv logs/LIVE-SESSION/CURRENT-SESSION.md logs/LIVE-SESSION/history/2026-02-15.md

# Start fresh
/save

“Context doesn’t match reality”

Issue: Session says “pending” but files are actually processed

Solution:

# Check actual state
/jarvis-briefing

# Update session with correct info
/save
# Provide accurate current state

Advanced: Session Hooks

Mega Brain includes session hooks that auto-save/restore:

• SessionStart hook - Offers to resume if session exists
• Stop hook - Reminds to save before closing

These run automatically. You can also manually /save anytime.

Session Persistence Tips

1. One Session Per Goal

Good:
  Session: "Process all Cole Gordon materials"
  Scope: 5 files, DNA extraction
  Duration: 2-3 days

Bad:
  Session: "Do Mega Brain stuff"
  Scope: Everything
  Duration: Forever

2. Archive Completed Sessions

# After completing a session
mv logs/LIVE-SESSION/CURRENT-SESSION.md logs/LIVE-SESSION/history/cole-gordon-complete.md

# Start fresh for next project
/save

3. Session Templates

For recurring workflows:

# Template: New Expert Processing

## Task
Process all materials for [EXPERT_NAME]

## Checklist
- [ ] Ingest all videos/docs
- [ ] Process each through pipeline
- [ ] Extract DNA (after 3+ sources)
- [ ] Review heuristics
- [ ] Update agent memories

## Files
- inbox/[EXPERT]/...

## Next Steps
1. /ingest [first video]
2. /process-jarvis [first file]
...

Next Steps

JARVIS Briefing

Check system state after resuming

Workflow Guide

Learn optimal processing workflows

Session Hooks

Understand auto-save/restore behavior

Logging System

Explore comprehensive logging