BMAD-METHOD/bmad-agent/tasks/validate-knowledge-base.md

86 lines
3.6 KiB
Markdown

# Task: Validate Knowledge Base
## Description
Systematically review and validate the project knowledge base to ensure completeness, consistency, and accuracy across all knowledge files and agent customizations.
## Input Required
- Knowledge files in `.ai` directory
- Agent configuration files
- Project documentation (PRD, architecture, etc.)
## Steps
1. **Knowledge File Completeness Check**
- Verify all required knowledge files exist:
- `.ai/project-context.md`
- `.ai/tech-stack.md`
- `.ai/data-models.md`
- `.ai/deployment-info.md`
- Check that each file follows its corresponding template structure
- Ensure all major sections in each file contain meaningful content
- Flag empty or placeholder sections as incomplete
2. **Internal Consistency Validation**
- Cross-reference information across knowledge files to identify:
- Terminology inconsistencies (different terms for same concept)
- Contradictory information (conflicting statements)
- Redundant information (same data in multiple places)
- Ensure references between files are accurate and up-to-date
- Verify versioning information is consistent if present
3. **External Consistency Validation**
- Compare knowledge files against authoritative project documents:
- Project Brief
- PRD
- Architecture documentation
- UI/UX specifications
- Identify any discrepancies or outdated information
- Flag potentially incorrect information for review
4. **Agent Customization Verification**
- Check that agent customization strings in configuration files:
- Accurately reflect current project knowledge
- Contain specific, actionable information
- Are consistent with knowledge file contents
- Verify all specialized agents have appropriate customizations
5. **Knowledge Gap Analysis**
- Identify missing information that should be documented
- Look for vague or imprecise statements that need clarification
- Note areas where more detailed documentation would be beneficial
- Check for outdated information that needs updating
6. **Create Validation Report**
- Summarize findings from all validation steps
- List specific issues categorized by severity:
- Critical: Inconsistencies that could lead to errors
- Important: Missing information needed for effective work
- Minor: Improvements that would enhance clarity
- Provide specific recommendations for each issue
- Generate actionable tasks to address knowledge gaps
## Output
A comprehensive validation report (`.ai/knowledge-validation-report.md`) containing:
- Overall knowledge base health assessment
- Specific issues identified across all knowledge files
- Consistency analysis between knowledge files and project documentation
- Agent customization review findings
- Prioritized list of recommendations to improve the knowledge base
## Knowledge Validation Checklist
- [ ] All required knowledge files exist and follow templates
- [ ] All major sections contain meaningful content
- [ ] Terminology is used consistently across files
- [ ] No contradictory information exists between files
- [ ] Information is consistent with authoritative project documents
- [ ] Agent customizations accurately reflect current knowledge
- [ ] Knowledge gaps are identified and documented
- [ ] Recommendations are specific and actionable
- [ ] Validation report is clear and comprehensive
## Validation Criteria
- All knowledge files have been reviewed
- Internal and external consistency has been verified
- Knowledge gaps are clearly documented
- Recommendations are prioritized by importance
- Report is structured for easy action by the team