Files
gh-unclecode-claude-code-to…/commands/pp-resume.md
2025-11-30 09:03:52 +08:00

124 lines
3.1 KiB
Markdown

# Resume Project Session
Start a new chat session with full project context loaded.
## Arguments
$ARGUMENTS
## Purpose
This is the FIRST command to run when starting a new chat session. It:
1. Loads all project context into memory
2. Understands current state
3. Prepares to continue work
4. Optionally focuses on specific direction from query
## Execution Instructions
### Step 1: Verify Project Exists
```bash
ls .context/project/INDEX.md
```
If not found, inform user: "No project structure found. Run `/pp-init` first."
### Step 2: Read Root Context
Read these files in order:
1. `.context/project/INDEX.md` - Project overview, subprojects, active subproject, high-level status
2. `.context/project/WORKFLOW.md` - How this system works
3. `.context/project/PRINCIPLES.md` - Methodology and rules
4. `.context/project/LESSONS.md` - Past learnings
### Step 3: Read Active Subproject Context
From INDEX.md, identify active subproject, then read:
1. `.context/project/{active}/STATUS.md`
2. `.context/project/{active}/TODO.md`
3. `.context/project/{active}/CODEBASE.md`
4. `.context/project/{active}/LESSONS.md`
5. `.context/project/{active}/PRINCIPLES.md` (if exists - overrides root)
### Step 4: Read Recent Changes
Read last 3-5 entries from:
- `.context/project/{active}/CHANGELOG.md`
### Step 5: Process Optional Query
If query provided:
- Parse for direction (e.g., "continue booking flow", "fix the API bug")
- Focus context on relevant TODO items
- Identify relevant LESSONS if debugging
### Step 6: Check for Next Session Instructions (NEW)
**After loading all standard files**, check for continuation instructions:
```bash
ls .context/project/{active}/NEXT.md
```
**If NEXT.md exists:**
1. Read `.context/project/{active}/NEXT.md`
2. Extract history file path from NEXT.md
3. Read that history file for previous session context
4. Note the full chat transcript path (but DO NOT load it)
**If NEXT.md does NOT exist:**
- Skip this step, proceed to summary
### Step 7: Generate Resume Summary
Output a compact summary:
```
## Session Resumed: {PROJECT_NAME}
**Active**: {subproject}/ - {current phase}
**Status**: {brief status}
{IF NEXT.md exists:}
### Instructions from Previous Session
{content from NEXT.md "What To Do Next" section}
**Previous Session**: {date} - {brief summary from history file}
### Current Focus
{from STATUS.md current focus}
### In Progress
{tasks marked [>] from TODO.md}
### Recent Changes
{last 2-3 changelog entries, one line each}
### Key Context
{if query provided, relevant context for that direction}
---
Ready to continue. {If NEXT.md exists: "Following previous session's plan." else: "What would you like to work on?"}
```
## Query Examples
```bash
# General resume - load everything
/pp-resume
# With direction - focus on specific area
/pp-resume continue the booking flow implementation
/pp-resume fix the WhatsApp routing bug
/pp-resume review yesterday's changes
```
## Important Notes
- This command is READ-ONLY (no file modifications)
- After this, Claude has full project context
- User can immediately start working
- If query mentions something not in TODO, suggest adding it