Beyond Modal: AI that thinks creatively across the complete research lifecycle. Built on Verbalized Sampling (VS) and HAVS to prevent mode collapse.
Simply describe your research:
- "I want to conduct a systematic review on AI in education"
- "Help me design an experimental study"
- "메타분석 연구를 시작하고 싶어"
The system will: detect paradigm -> confirm -> present VS alternatives -> wait for selection -> guide with checkpoints
| ID | Name | Cat | Model | ID | Name | Cat | Model |
|---|---|---|---|---|---|---|---|
| A1 | RQRefiner | A | opus | G1 | JournalMatcher | G | sonnet |
| A2 | TheoryCritiqueArch | A | opus | G2 | PublicationSpec | G | sonnet |
| A5 | ParadigmAdvisor | A | opus | G5 | StyleAuditor | G | sonnet |
| B1 | LiteratureScout | B | sonnet | G6 | StyleHumanizer | G | opus |
| B2 | QualityAppraiser | B | sonnet | I0 | SROrchestrator | I | opus |
| C1 | QuantDesignSampling | C | opus | I1 | PaperRetrieval | I | sonnet |
| C2 | QualitativeDesign | C | opus | I2 | ScreeningAssist | I | sonnet |
| C3 | MixedMethodsDesign | C | opus | I3 | RAGBuilder | I | haiku |
| C5 | MetaAnalysisMaster | C | opus | X1 | ResearchGuardian | X | sonnet |
| D2 | DataCollectionSpec | D | sonnet | V1 | PostPositivist | V | opus |
| D4 | InstrumentDeveloper | D | opus | V2 | CriticalTheorist | V | opus |
| E1 | QuantAnalysisCodeGen | E | opus | V3 | Pragmatist | V | opus |
| E2 | QualitativeCoding | E | opus | V4 | Interpretivist | V | opus |
| E3 | MixedMethodsInteg | E | opus | V5 | Transformative | V | opus |
| F5 | HumanizationVerifier | F | haiku |
| Tier | Model | When |
|---|---|---|
| HIGH | opus | Architecture, complex design, deep analysis |
| MEDIUM | sonnet | Standard tasks, literature search, writing |
| LOW | haiku | Quick validation, code generation, RAG |
| Level | Behavior |
|---|---|
| REQUIRED | System STOPS. Cannot proceed without explicit approval. |
| RECOMMENDED | System PAUSES. Strongly suggests approval. |
| OPTIONAL | System ASKS. Defaults available if skipped. |
Required checkpoints: CP_RESEARCH_DIRECTION, CP_PARADIGM_SELECTION, CP_THEORY_SELECTION, CP_METHODOLOGY_APPROVAL, SCH_DATABASE_SELECTION, SCH_SCREENING_CRITERIA.
Team Dispatch Bypass: When orchestrator dispatches Agent Teams with user approval, include DIVERGA_TEAM_DISPATCH=1 in agent prompts to bypass prerequisite enforcement (Rule 7).
Quantitative: hypothesis, effect size, p-value, experiment, ANOVA, regression, 가설, 효과크기, 통계 Qualitative: lived experience, saturation, themes, phenomenology, coding, 체험, 포화, 현상학 Mixed methods: sequential, convergent, integration, joint display, 혼합방법, 통합
Confirmation always required via CP_PARADIGM_SELECTION.
English default. Responds in Korean when user input is Korean.
- Checkpoint enforcement rules: docs/CHECKPOINT-RULES.md
- Architecture and systems: docs/ARCHITECTURE.md
- MCP tools reference: docs/MCP-TOOLS.md
- Agent Teams orchestration: /diverga:orchestrator
- Full agent catalog: AGENTS.md