Back to Agents

Category A: Foundation Agents

3 agents for theoretical foundations and research excellence

Foundation agents establish the theoretical and methodological groundwork for your research. Using Verbalized Sampling (VS) methodology, they prevent mode collapse by presenting creative alternatives across the typicality spectrum.

Core Principle

Prevent mode collapse through creative theory selection, built-in critique, and paradigm alignment

Research Question Refiner

A1

Transform vague research ideas into precise, answerable research questions using FINER/PICO/SPIDER frameworks

HIGH • Opus
Full VS 5-Phase
CP_RESEARCH_DIRECTION

Capabilities

  • •FINER criteria validation (Feasible, Interesting, Novel, Ethical, Relevant)
  • •PICO framework for quantitative questions
  • •SPIDER framework for qualitative questions
  • •Scope refinement to prevent overly broad/narrow questions
  • •VS-driven alternative question generation

VS Process

Identifies modal question patterns, generates 3 alternatives with T-Scores (0.6, 0.4, <0.3), presents for human selection

Example

Input
"How does AI affect education?"
Output
Direction A (T≈0.6): "How do AI chatbots improve speaking proficiency in EFL classrooms?" | Direction B (T≈0.4): "What metacognitive strategies emerge when learners interact with AI feedback systems?" | Direction C (T<0.3): "How does neuroplasticity theory explain learning differences in AI-augmented vs. traditional environments?"
Triggers:research questionRQrefine questionresearch problem

Theory & Critique Architect

A2

Design theoretical frameworks, generate self-critique, and visualize conceptual models — combining theory construction, devil's advocacy, and framework visualization

HIGH • Opus
Full VS 5-Phase
CP_THEORY_SELECTION

Capabilities

  • •Modal theory identification (T > 0.8)
  • •Long-tail theory exploration (T < 0.4)
  • •Theory integration and hybrid frameworks
  • •Cross-domain theory adaptation
  • •Hypothesis derivation from theory
  • •Built-in self-critique and devil's advocacy (formerly A3)
  • •Anticipate "Reviewer 2" criticisms and competing hypotheses
  • •Conceptual framework visualization via Mermaid diagrams (formerly A6)
  • •Variable relationship mapping and mediation/moderation models

VS Process

Stage 1: Identify modal theories (TAM, UTAUT, etc.) | Stage 2: Generate differentiated alternatives | Stage 3: Self-critique each option | Stage 4: Present 3 options with T-Scores | Stage 5: Human selection | Stage 6: Framework elaboration and visualization

Example

Input
"AI adoption in education"
Output
Modal (T=0.92): TAM | Direction A (T≈0.6): Self-Determination Theory x TAM integration | Direction B (T≈0.4): Cognitive Load Theory + Adaptive Ecosystem | Direction C (T<0.3): Neuroplasticity-based technology learning framework | Self-critique: Direction A risks superficial integration...
Triggers:theoretical frameworktheoryconceptual modeltheoretical foundationcritiquevisualize framework

Paradigm & Worldview Advisor

A5

Guide researchers through ontological, epistemological, and axiological foundations

HIGH • Opus
Standard
CP_PARADIGM_SELECTION

Capabilities

  • •Paradigm identification (Positivist, Interpretivist, Pragmatic, Transformative)
  • •Ontological positioning (Realism vs. Relativism)
  • •Epistemological stance (Objectivism vs. Constructivism)
  • •Axiological considerations (Value-free vs. Value-laden)
  • •Methodology-paradigm alignment check

VS Process

Ask foundational questions | Present paradigm options | Explain implications for methods | Guide paradigm selection | Ensure consistency

Example

Input
"Should I use qual or quant methods?"
Output
First clarify: What is the nature of reality you're studying? (Ontology) | How can knowledge about it be acquired? (Epistemology) | Then recommend: Positivist → Quant, Interpretivist → Qual, Pragmatic → Mixed
Triggers:paradigmontologyepistemologyworldviewphilosophical foundationsmethodology alignment

Checkpoint Integration

Foundation agents enforce critical decision points:

CP_RESEARCH_DIRECTION

A1

Research question finalized, VS alternatives presented

CP_THEORY_SELECTION

A2

Theoretical framework chosen from VS options (includes critique and visualization)

CP_PARADIGM_SELECTION

A5

Paradigm confirmed (Quant/Qual/Mixed)

Verbalized Sampling (VS) in Action

Foundation agents use VS to prevent mode collapse:

1

Modal Awareness

Identify predictable recommendations (T > 0.8)

2

Divergent Exploration

Generate alternatives across T-Score spectrum

3

Human Selection

Present options, WAIT for decision

4

Elaboration

Develop selected direction

5

Validation

Ensure defensibility and rigor

Typical Research Foundation Workflow

1
A1

Refine research question

CP_RESEARCH_DIRECTION
2
A5

Select paradigm

CP_PARADIGM_SELECTION
3
A2

Develop theory, self-critique, and visualize framework

CP_THEORY_SELECTION

Build Strong Research Foundations

Start with Category A agents to establish theoretical and ethical rigor.

Explore Category B: Evidence