shield_lock
JinDaGe
← Back to skills
extension
Category: AI Agent CapabilitiesNo API key required

convo

Complete conversation lifecycle management - start (understanding & alignment), capture (mid-conversation context), and exit (debrief & improvement). Use automatically at conversation boundaries and checkpoints.

personAuthor: jakexiaohubgithub
open_in_newRepositorydownloadDownload package

Conversation Lifecycle

Manages the full conversation lifecycle: start, capture, and exit.

Overview

| Phase | When | Purpose | |-------|------|---------| | Start | Beginning of conversation | Establish understanding and alignment | | Capture | Mid-conversation | Save context for continuity | | Exit | End of conversation | Debrief, feedback, and improvement |

Part 1: Start

Establishes understanding and alignment before diving into execution.

When to Start

  • Automatically at the start of every new conversation
  • When user explicitly says "convo-start"
  • When topic shifts significantly mid-conversation

Phase 1: Understanding

Actions:

  1. Read the request carefully
  2. Identify:
    • What type of work is this? (feature, spec review, planning, research, fix)
    • Which agents/patterns are relevant?
    • What context do I need?
  3. Gather initial context (follow research-priority pattern)

Research Priority Pattern:

  1. Repo docs (CLAUDE.md, README, existing specs)
  2. Linear/issue tracker (if referenced)
  3. Ask user for clarification
  4. External sources (last resort)

Phase 2: Alignment

1. State Your Understanding

I understand you want to [specific goal].

This involves:
- [Scope item 1]
- [Scope item 2]

2. Propose Execution Plan

Here's my approach:
1. [Step 1]
2. [Step 2]
3. [Step 3]

I'll use:
- [Relevant agent/pattern]
- [Another relevant pattern]

3. Wait for Confirmation

Do NOT proceed until user confirms or corrects the plan.

Quick Templates

For simple tasks:

Got it - you want [X]. I'll [approach]. Sound good?

For complex tasks:

Let me make sure I understand:

**Goal**: [What you want to achieve]
**Scope**: [What's included/excluded]
**Approach**: [How I'll tackle it]
**Output**: [What you'll get]

Anything to adjust before I start?

When to Skip Full Alignment

Skip when:

  • Simple question requiring quick answer
  • User explicitly says "just do X"
  • Continuing prior conversation with established context

Never skip when:

  • Creating or modifying documents/code
  • Work will take significant time
  • Multiple approaches are possible
  • User's intent is unclear

Part 2: Capture

Saves context mid-conversation for continuity and handoff.

When to Capture

Automatic Triggers

  1. Significant decision made - Architecture choice, approach selected, scope change
  2. Large work completed - Document created, major code changes, phase finished
  3. Topic shift detected - Before switching to new topic
  4. Natural breakpoint - Before user goes to meeting, logical pause

On Request

  • User asks "save where we are"
  • User needs to leave but wants to continue later
  • Handoff to another person/session

What to Capture

| Element | Why It Matters | |---------|----------------| | Current state | Where we are right now | | Decisions made | What was decided and why | | Files changed | What was created/modified | | Next steps | What comes next |

Capture Formats

Inline Capture (Quick)

---
**Checkpoint**: [Brief description]

**Status**: [Current state of work]
**Just completed**: [What was done]
**Next**: [What's queued]
---

Context File (Detailed)

Create convo-captures/YYYY-MM-DD-HHMM-topic.md:

# Context Capture: [Topic]

**Date**: YYYY-MM-DD HH:MM
**Status**: In Progress / Paused / Blocked

## Current State
[Where we are in the work]

## Decisions Made
| Decision | Rationale | Alternatives Considered |
|----------|-----------|------------------------|
| [Choice] | [Why] | [Other options] |

## Files Changed
**Created**: [list]
**Modified**: [list]

## Next Steps
1. [ ] [Immediate next step]
2. [ ] [Following step]

## To Resume
[Instructions for picking this back up]

Part 3: Exit

Wraps up conversations with debrief, feedback, and knowledge capture.

Triggers

  • User says "we're done", "that's it", "looks good"
  • Conversation naturally concluding
  • User explicitly requests exit/debrief
  • Topic drift suggests starting fresh

Exit Protocol

Step 1: Review

Let me review our conversation:

**Original goal**: [State the goal]

**What we accomplished**:
- [Outcome 1]
- [Outcome 2]

**Files created/modified**:
- `path/file.md` - [Brief description]

Step 2: Tasks Remaining

**Tasks Remaining**:

For this effort:
- [ ] [Remaining task 1]
- [ ] [Remaining task 2]

Step 3: Ask for Feedback

**Your feedback**:

1. What worked well in this conversation?
2. What would you have liked to see improved?
3. Were there moments where I was unclear or off-track?

Step 4: Propose Changes

After receiving feedback:

**Based on our conversation, I propose these changes**:

1. [File path]: [Specific change]
2. [File path]: [Specific change]

Should I make these changes?

Step 5: Create Convo-Exit Document

File: convo-exits/YYYY-MM-DD-HHMM-topic.md

# Conversation Exit: [Topic]

**Date**: YYYY-MM-DD HH:MM
**Repo**: [which repo(s) were involved]

## Original Goal
[What we set out to do]

## Outcomes
**Files Created**: [list]
**Files Modified**: [list]
**Decisions Made**: [list]

## User Feedback
**What worked well**: [feedback]
**What could improve**: [feedback]

## Improvements Made
- [ ] [Change 1] - [Status]

## Next Steps
**Immediate**: [list]
**Future**: [list]

Step 6: Activity Log Update

If brain repo has personal/activity-log.md AND significant work was done:

## YYYY-MM-DD HH:MM - [Topic]

**What I accomplished**:
- [Accomplishment 1]
- [Accomplishment 2]

**Repos affected**: [list]
**Related**: `convo-exits/YYYY-MM-DD-HHMM-topic.md`

Anti-patterns

| Wrong | Right | |-------|-------| | Immediately starting work | State understanding first, get confirmation | | Acting on assumptions | Explicitly state assumptions, ask for confirmation | | No record of decisions | Capture decisions with rationale | | Context dump without resume instructions | Clear "To continue, do X" instructions | | Skipping debrief | Always offer exit protocol for significant work |

Success Metrics

Start

  • [ ] User never confused about what you're doing
  • [ ] Plan is specific enough to evaluate
  • [ ] User can correct misunderstandings early

Capture

  • [ ] Can resume work without context loss
  • [ ] Can hand off to someone else smoothly
  • [ ] Past decisions are understandable

Exit

  • [ ] Clear record of what was accomplished
  • [ ] Learnings captured for future
  • [ ] Improvements made to the system
  • [ ] User feels heard and in control