Research Architect Literature

Research Architect Literature and Exemplar Analysis

Use this branch in two passes when target references exist, or once after brainstorming when they do not.

Purpose

Turn target papers and surrounding literature into:

• an explanation of how the target paper works;
• a functional account of its strengths;
• a design-family classification;
• a project-specific adaptation plan;
• a map of concepts, methods, comparisons, evidence standards, debates, and gaps;
• candidate research spines grounded in feasible evidence or warrants.

The output should teach the user what to do and why each step is necessary.

Pass 1 — Target-reference analysis

Run before ordinary brainstorming when project_config.json contains target_references.

Read:

paper_output/project_config.json
paper_output/source_inventory.md
target reference files

Create:

paper_output/exemplar_logic_profile.md
paper_output/exemplar_adaptation_plan.md

At this stage, the adaptation plan may contain provisional user-project analogues and explicit questions that materially affect the design.

Pass 2 — Literature positioning and final adaptation

Run after candidate questions exist.

Read:

paper_output/project_config.json
paper_output/source_inventory.md
paper_output/idea_candidate_matrix.md
paper_output/research_question_options.md
paper_output/exemplar_logic_profile.md        # when target references exist
paper_output/exemplar_adaptation_plan.md      # preliminary version when available

Source roles

Assign each source one or more roles:

| Role | Purpose | |---|---| | Primary exemplar | Supplies the overall research logic the user wants to learn from | | Secondary exemplar | Supplies a particular design, analysis, display, or rhetorical strength | | Concept source | Defines the phenomenon, construct, or task | | Theory source | Defines or motivates a theoretical or conceptual orientation | | Method source | Provides data-generation, analysis, or interpretive procedures | | Comparison/warrant source | Provides baseline, counterfactual, negative-case, rival-explanation, or argumentative logic | | Benchmark/data source | Provides a dataset, corpus, protocol, instrument, archive, or measurement norm | | Citation source | Supports a specific literature claim after separate verification |

Required outputs

Create:

paper_output/reference_materials/source_index.md
paper_output/concept_dossier.md
paper_output/corpus_inventory.md
paper_output/exemplar_method_map.md
paper_output/exemplar_logic_profile.md        # when target references exist
paper_output/exemplar_adaptation_plan.md      # when target references exist
paper_output/sota_gap_map.md
paper_output/motivation_options_after_literature.md

Exemplar logic profile

Use templates/exemplar_logic_profile.md. Analyze:

• research object and starting tension;
• theory or conceptual role;
• research-question move;
• contribution type;
• unit, participant, case, site, document, corpus, or sample-selection logic;
• evidence generation;
• analysis or interpretation logic;
• comparison, counterfactual, rival explanation, negative case, or counterargument;
• validity and credibility logic;
• result and evidence-display sequence;
• rhetorical sequence;
• limitation and inference-boundary strategy;
• context-dependent assumptions;
• transferable strengths;
• copying boundaries.

Explain why each load-bearing move works.

Adaptation plan

Use templates/exemplar_adaptation_plan.md.

| Transfer ID | Reference Move | Why It Works | Transferable Principle | User-Project Analogue | Required Evidence or Material | Decision | Context Difference | Feasibility | Copying Boundary | Next Action | |---|---|---|---|---|---|---|---|---|---|---|

Allowed decisions:

• TRANSFER: the principle fits with only project-specific implementation;
• ADAPT: the function transfers and the design changes materially;
• REPLACE: the reference move is incompatible and another move serves the same function;
• OMIT_WITH_RATIONALE: the function is unnecessary for the user's bounded claim.

Every load-bearing reference move receives a decision. Topic substitution without design reasoning fails the adaptation.

Gap map schema

| Gap ID | Candidate Gap or Tension | What Prior Work Establishes | What Remains Unresolved | Required Evidence or Warrant | User Feasibility | Possible Claim | Design Family | Risk | |---|---|---|---|---|---|---|---|---|

Exemplar method map schema

| Exemplar ID | Why Included | Research Spine | Logic of Inquiry | Evidence Sequence | Evidence-Display Functions | Comparison or Warrant Logic | Reusable Structural Lesson | Context Boundary | Copying Risk | |---|---|---|---|---|---|---|---|---|---|

Motivation options

Generate 2-5 candidate research spines:

| Option | One-Sentence Research Spine | Reference Logic Preserved | Required Adaptation | User-Specific Contribution | Required Evidence or Warrant | Why It Matters | Why It Is Feasible | Main Risk | Scope Boundary | |---|---|---|---|---|---|---|---|---|---|

Stop for the user to choose, modify, or reject the options. After confirmation, write:

paper_output/confirmed_research_spine.md

Rules

• Keep exemplar learning separate from citation support.
• Preserve source-specific claims, wording, data, findings, figures, quotation choices, and citation choices as source-specific.
• Infer the design family from the target paper and project; record uncertainty when hybrid or ambiguous.
• Verify that prior work leaves the proposed gap or tension unresolved.
• Prefer an adaptation the user can execute over a superficial imitation of a resource-intensive design.
• Mark uncertain source claims as needs_verification.
• Record contradictory sources and rival interpretations.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: mmTheBest
• Source: mmTheBest/Research-Architect
• License: MIT

Install and usage instructions live in the source repository linked above.