session-handoff
Session Handoff Framework
When This Activates
This skill activates when:
- Starting a new session with continuity expectations
- User explicitly asks about previous work
- Context from last session is needed
Session Continuity Data
The system automatically tracks:
Last Session Summary
<session-continuity>
[LAST SESSION] Description of what was worked on
Files: list of key files
</session-continuity>
Recent Decisions
Important choices made recently that affect current work.
Learned Patterns
- Gotchas discovered
- Decisions recorded
- Patterns established
Handoff Protocol
1. Acknowledge State (What we know)
"Continuing from the last session where we worked on [X]...
Current state:
- Last worked on: [from session-continuity]
- Pending items: [from PENDING WORK]
- Blockers: [from BLOCKERS if any]"
2. Address Blockers First (If any)
"I see there's a blocker we need to address first:
[blocker description]
Let me investigate this before continuing..."
3. Propose Next Actions
"Based on our pending work, I suggest we:
1. [Most important/blocked item first]
2. [Next priority item]
3. [Continue from where we left off]
Which would you like to tackle?"
4. Check for External Changes
"Let me check what changed since then..."
→ Use git_awareness for code changes
→ Check for new errors or issues
MCP Tools for Continuity
# Get session context
memory_sessions list_sessions=true
# Get recent decisions
memory_sessions category=decision limit=5
# Check what changed
memory_query "recent changes to [area]"
Session Recovery Scenarios
"What were we working on?"
- Check
<session-continuity>injection - Query recent session observations
- Summarize last activities
"Continue from last time"
- Recall last session context
- Check git for any external changes
- Resume with state awareness
"Pick up [specific task]"
- Search session history for task
- Load relevant context
- Resume from last known state
State Persistence
Sessions are persisted in:
~/.claude-dash/sessions/
├── {project}/
│ └── session-{timestamp}.json
├── observations.json (cross-session learnings)
├── summaries/ (per-project summaries)
└── transcripts/ (compressed conversation logs)
Best Practices
- End sessions cleanly - Summarize what was done
- Note blocking issues - Record what couldn't be resolved
- Mark next steps - Clarify intended continuation
- Save decisions - Record important choices with context
Actionable Handoff Checklist
At session start:
- Read
<session-continuity>injection - Check for
[PENDING WORK]items - Check for
[BLOCKERS]- address these first - Acknowledge state to user
- Propose next actions based on priorities
At session end:
- Summarize what was completed
- Note any new blockers discovered
- Update roadmap if tasks completed
- Record any important decisions made
Continuity Phrases
Use these to acknowledge context:
- "Picking up from where we left off..."
- "Based on our last session where we..."
- "Continuing the work on [X]..."
- "As we discussed previously..."
Handling Missing Context
If session context isn't available:
"I don't have specific context from our last session.
Could you remind me what we were working on, or should
I check the recent git history and project state?"
More from jamelna-apps/claude-dash
cost-tracking
When user mentions "spending", "usage", "tokens", "API cost", "budget", "expensive", or wants to understand Claude API costs. Provides cost awareness and optimization guidance.
11page-cro
When the user mentions "conversion", "CRO", "landing page", "not converting", "bounce rate", "optimize page", or asks about improving page performance.
7error-diagnosis
When user encounters "error", "exception", "failed", "stack trace", "crashed", or needs error categorization. Provides structured root cause analysis and prevention strategies.
4testing-strategy
When the user mentions "test", "testing", "unit test", "integration", "e2e", "coverage", "TDD", "mock", or asks about testing approaches.
3mode-awareness
When explicitly switching modes or when mode detection shows low confidence. Provides guidance on optimal approach for each detected mode (debugging, performance, feature, refactor, exploration, infrastructure).
3launch-strategy
When the user mentions "launch", "release", "announcement", "go-to-market", "GTM", "product launch", or asks about launching a product or feature.
3