Document use of subagents

[cmungall]

• Claude: https://docs.anthropic.com/en/docs/claude-code/sub-agents
• Goose: https://block.github.io/goose/docs/experimental/subagents/

[cmungall]

Q from @matentzn on slack:

I had one Q about the subagents: why did you divide them by role rather than workflow? So having a metadata and a reasoner agent instead of a new term request agent and metadata change review agent etc?

My answer: we need both

[cmungall]

@claude do a deep dive on this, look at how they are used in mondo https://github.com/monarch-initiative/mondo/tree/master/.claude

Write a doc that is aimed at people wanting to use subagents for 'ontology coding' (read the docs in this repo if you don't know what I mean). You can make the doc a kind of case study with mondo

[claude]

Claude finished @cmungall's task —— View job • claude/issue-13-20250816-0135 • Create PR ➔

---

Analysis Complete: Subagents for Ontology Coding Documentation

I've completed a comprehensive deep dive into subagent usage for ontology coding, creating detailed documentation based on the MONDO project as a case study.

Todo List:

• Research this repo's documentation to understand "ontology coding"
• Examine Mondo's .claude directory structure and subagent configuration
• Research the provided Claude Code and Goose subagent documentation
• Analyze subagent organization patterns (role-based vs workflow-based)
• Create comprehensive documentation as a Mondo case study
• Review and finalize documentation

Key Findings & Documentation Created

📋 MONDO's Role-Based Architecture

The MONDO project uses 6 specialized subagents organized hierarchically:

• task-coordinator: Master orchestrator (must be used first)
• deep-research-specialist: Literature research and evidence synthesis
• design-pattern-advisor: Ontology pattern identification and application
• identifier-validator: External identifier verification
• metadata-checker: Compliance and quality assurance
• ontology-reasoner: Logical consistency validation

🎯 Role vs. Workflow Organization

The analysis confirms that both approaches are needed, but MONDO demonstrates that role-based organization with workflow coordination provides:

• Composability: Mix and match agents for different workflows
• Expertise Concentration: Deep specialization in each domain
• Quality Assurance: Systematic validation at every step
• Maintainability: Clear separation of concerns

📚 Comprehensive Tutorial Created

I've created docs/tutorials/subagents-for-ontology-coding.md covering:

• Background on subagents and their benefits for ontology work
• Detailed analysis of each MONDO subagent and its role
• Why role-based organization works for ontology coding
• Practical implementation guidance for other projects
• Integration with ODK workflows and GitHub Actions

The documentation is targeted at ontology coding practitioners and uses MONDO as a comprehensive case study, exactly as requested.

🔧 Files Modified

• Created: docs/tutorials/subagents-for-ontology-coding.md (comprehensive tutorial)
• Updated: mkdocs.yml (added new tutorial to navigation)