BMAD-METHOD/bmad-core/tasks/session-kickoff.md

219 lines
6.5 KiB
Markdown

# Session Kickoff
This task ensures AI agents have complete project context and understanding before starting work. It provides systematic session initialization across all agent types.
## Purpose
- Establish comprehensive project understanding
- Validate documentation consistency
- Identify current project state and priorities
- Recommend next steps based on evidence
- Prevent context gaps that lead to suboptimal decisions
## Process
### 1. Memory Bank Review (Primary Context)
**Priority Order**:
1. **Memory Bank Files** (if they exist): `docs/memory-bank/`
- `projectbrief.md` - Project foundation and scope
- `activeContext.md` - Current work and immediate priorities
- `progress.md` - Project state and completed features
- `systemPatterns.md` - Architecture and technical decisions
- `techContext.md` - Technology stack and constraints
- `productContext.md` - Problem space and user needs
**Analysis Required**:
- When were these last updated?
- Is information current and accurate?
- Any apparent inconsistencies between files?
### 2. Architecture Documentation Review
**Primary References** (check which exists):
- `/docs/architecture.md` - General backend/system architecture (greenfield)
- `/docs/brownfield-architecture.md` - Enhancement architecture for existing systems
- `/docs/frontend-architecture.md` - Frontend-specific architecture
- `/docs/fullstack-architecture.md` - Complete full-stack architecture
**Key Elements to Review**:
- Core architectural decisions and patterns
- System design and component relationships
- Technology choices and constraints
- Integration points and data flows
- API documentation
- Database schemas
### 3. Development History Review
**Recent Dev Journals**: `docs/devJournal/`
- Read last 3-5 entries to understand recent work
- Identify patterns in challenges and decisions
- Note any unresolved issues or technical debt
- Understand development velocity and blockers
**Current ADRs**: `docs/adr/`
- Review recent architectural decisions
- Check for pending or superseded decisions
- Validate alignment with current architecture
- Skip archived ADRs (consolidated in architecture docs)
### 4. Project Documentation Scan
**Core Documentation**:
- `README.md` - Project overview and setup
- `CHANGELOG.md` - Recent changes and releases
- Package manifests (`package.json`, `requirements.txt`, etc.)
- Configuration files
**Additional Context**:
- Issue trackers or project boards
- Recent commits and branches
- Test results and coverage reports
### 5. Current State Assessment
**Development Environment**:
```bash
# Check git status
git status
git log --oneline -10
# Check current branch and commits
git branch -v
# Review recent changes
git diff --name-status HEAD~5
```
**Project Health**:
- Are there failing tests or builds?
- Any urgent issues or blockers?
- Current sprint/iteration status
- Outstanding pull requests
### 6. Consistency Validation
**Cross-Reference Checks**:
- Does Memory Bank align with actual codebase?
- Are ADRs reflected in current architecture?
- Do dev journals match git history?
- Is documentation current with recent changes?
**Identify Gaps**:
- Missing or outdated documentation
- Undocumented architectural decisions
- Inconsistencies between sources
- Knowledge gaps requiring clarification
### 7. Agent-Specific Context
**For Architect Agent**:
- Focus on architectural decisions and system design
- Review technical debt and improvement opportunities
- Assess scalability and performance considerations
**For Developer Agent**:
- Focus on current work items and immediate tasks
- Review recent implementation patterns
- Understand testing and deployment processes
**For Product Owner Agent**:
- Focus on requirements and user stories
- Review product roadmap and priorities
- Assess feature completion and user feedback
### 8. Next Steps Recommendation
**Based on Evidence**:
- What are the most urgent priorities?
- Are there any blockers or dependencies?
- What documentation needs updating?
- What architectural decisions are pending?
**Recommended Actions**:
1. **Immediate Tasks** - Ready to start now
2. **Dependency Resolution** - What needs clarification
3. **Documentation Updates** - What needs to be updated
4. **Strategic Items** - Longer-term considerations
## Quality Checklist
- [ ] Memory Bank reviewed (or noted if missing)
- [ ] Architecture documentation understood
- [ ] Recent development history reviewed
- [ ] Current project state assessed
- [ ] Documentation inconsistencies identified
- [ ] Agent-specific context established
- [ ] Next steps clearly recommended
- [ ] Any urgent issues flagged
## Output Template
```markdown
# Session Kickoff Summary
## Project Understanding
- **Project**: [Name and core purpose]
- **Current Phase**: [Development stage]
- **Last Updated**: [When Memory Bank was last updated]
## Documentation Health
- **Memory Bank**: [Exists/Missing/Outdated]
- **Architecture Docs**: [Current/Needs Update]
- **Dev Journals**: [Last entry date]
- **ADRs**: [Recent decisions noted]
## Current State
- **Active Branch**: [Git branch]
- **Recent Work**: [Summary from dev journals]
- **Project Health**: [Green/Yellow/Red with reasons]
- **Immediate Blockers**: [Any urgent issues]
## Inconsistencies Found
[List any documentation inconsistencies or gaps]
## Agent-Specific Context
[Relevant context for current agent role]
## Recommended Next Steps
1. [Most urgent priority]
2. [Secondary priority]
3. [Documentation updates needed]
```
## Integration Points
This task integrates with:
- **Memory Bank**: Primary source of project context
- **All Agents**: Universal session initialization
- **Document Project**: Can trigger if documentation missing
- **Update Memory Bank**: Can trigger if information outdated
- **Agent Activation**: Called at start of agent sessions
## Usage Patterns
**New Agent Session**:
1. Agent activates
2. Runs `session-kickoff` task
3. Reviews output and confirms understanding
4. Proceeds with informed context
**Project Handoff**:
1. New team member or AI session
2. Runs comprehensive kickoff
3. Identifies knowledge gaps
4. Updates documentation as needed
**Quality Gate**:
1. Before major feature work
2. After significant time gap
3. When context seems incomplete
4. As part of regular project health checks
## Notes
- This task should be lightweight for daily use but comprehensive for major handoffs
- Adapt depth based on project complexity and available time
- Can be automated as part of agent startup routines
- Helps prevent tunnel vision and context loss