rq-architect
Skill0 starsChanged 7 months agoMITView on GitHub
Architects research questions from problematization to operational quality. Combines Sandberg & Alvesson's (2010) problematization logic with FINERMAPS quality criteria and RQ-type taxonomy. Use whenever the user mentions RQ development, "the gap", dissertation framing, theoretical contribution, literature review positioning, A-journal submissions, or needs to go from a broad topic to a focused RQ. Trigger on phrases like "is my RQ strong enough", "how do I frame my contribution", "what's interesting about X", "help me develop a research question", "check my RQ", "what type of question is this", or when user uploads a dissertation chapter or paper. Also trigger for "I have a topic but no question yet", "build my RQ from scratch", or "develop hypothesis from my RQ".
In this repository19 other files it gives its agents.
- a-journal-architectureSkill
- citation-risk-auditorSkill
- desk-rejection-preventerSkill
- hallucination-auditSkill
- paper-revision-diagnosticSkill
- saturation-auditorSkill
- abductive-bridgeSkill
- argument-architectureSkill
- methods-narrative-architectSkill
- qualitative-coderSkill
- theory-map-architectSkill
- theory-section-architectSkill
- anti-ai-writingSkill
- data-integrity-guardSkill
- draft-review-loopSkill
- evidence-disciplineSkill
- non-sycophantSkill
- paper-feedback-pipelineSkill
- research-lifecycleSkill
The file
We split this skill into the rules it gives an agent
12 rules. About 0 are common, 12 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈12
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 35
When the user uploads a dissertation chapter or paper: read it fully before proceeding.
- Distinctive: line 92
Write out at least 3, ideally 5.
- Distinctive: line 115
The Third-Category Move: When a field operates with two exhaustive categories, ask whether the phenomenon under study requires a third …
- Distinctive: line 219
Do the RQs collectively challenge the framework, or operate within it?
- Distinctive: line 241
Verify that the overarching research problem, the review RQ (if present), and the empirical RQs share a single problematization logic.
- Distinctive: line 251
Ask: What is the phenomenon?
- Distinctive: line 266
Before proceeding, diagnose whether the forming question falls into common failure patterns:
- Distinctive: line 283
Write the question as completely as possible.
- Distinctive: line 297
If the RQ passes FINERMAPS but remains gap-spotting, proceed to Mode B (B1–B4) to push toward genuine problematization.
- Distinctive: line 364
Remove hedge phrases that dilute the theoretical stake: "explore", "examine", "investigate", "shed light on"
- Distinctive: line 365
Ensure the phenomenon appears in the RQ as something that *tests* the theory, not something the theory *explains*
- Distinctive: line 390
Before finalising any RQ:
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Research Question Architect
Two-layer architecture:
Layer 1 — Theoretical Ambition (Sandberg & Alvesson, 2010): Moves RQs from gap-spotting to genuine problematization. This is where intellectual contribution lives.
Layer 2 — Operational Quality (FINERMAPS + RQ-Type): Ensures the RQ is actually executable, correctly typed, and methodologically coherent. A theoretically sharp RQ that fails feasibility or ethics is still useless.
Core principle: Gap-spotting fills empty space in existing literature. Problematization challenges the assumptions that created that space. Only the latter produces interesting, influential theory.
Core warning: Naming an assumption is not the same as challenging it. A paper can identify a "convergence assumption", label it as a scope condition, and then proceed to operationalise it in all three research questions. This is the most common form of pseudo-problematization — and the hardest to detect.
Step 0: Detect Input Mode
| Input type | Go to |
| Existing RQ, research issue, or gap statement | Mode A: Diagnose & Upgrade |
| Topic / field / theories, no RQ yet | Mode B: Build from Scratch (theoretical) |
| Vague feeling / observation / empirical puzzle | Mode C: From Signal to RQ |
| Literature review chapter needing its own RQ | Mode D: Review RQ |
| Full suite of RQs needing coherent reformulation | Mode E: RQ Suite |
| Broad topic, needs stepwise operational development | Mode F: Stepwise Build |
| Finalised RQ needing quality audit | Mode G: FINERMAPS Check |
| Finalised RQ needing Hypothesis formulation | Mode H: Hypothesis Bridge |
| Any of the above + "is this strong enough?" | Mode A first, then relevant mode |
When the user uploads a dissertation chapter or paper: read it fully before proceeding. Extract existing RQs, identify the theoretical framework, and list the assumptions it leaves untouched before producing any output.
Mode A: Diagnose & Upgrade an Existing RQ
A1 — Gap-Spotting Diagnosis
Classify the existing RQ against Sandberg & Alvesson's typology:
Track-bound (gap-spotting):
- Confusion spotting — competing explanations exist, RQ tries to resolve them
- Neglect spotting — overlooked area / under-researched / lacks empirical support
- Application spotting — extends an existing theory into a new domain
Mixed:
- Critical confrontation — attacks a competing position from an alternative one
- Quasi-problematization — claims to challenge assumptions but imports a pre-fab answer
- New idea — sidesteps existing literature with an original concept
Disruptive:
- Genuine problematization — challenges assumptions underlying existing theory, including the researcher's own
Be direct. Name the category. Quote the gap-spotting rhetoric where present ("address this gap", "extend this literature", "under-researched area", "under what conditions").
⚠️ Assumption-Labelling without Assumption-Challenging: The subtlest anti-pattern. The researcher correctly identifies a foundational assumption, explains why it breaks down for their population — and then proceeds to use the theory built on that assumption as their primary analytical lens. The diagnosis is sharp; the RQs reproduce exactly what was diagnosed. Always check: do the actual research questions operate within the framework whose assumptions were just identified as problematic?
⚠️ Critical vocabulary ≠ problematization. Foucauldian, poststructuralist, or feminist framing applied to a neglect gap is still application spotting.
A2 — RQ Type Diagnosis
Classify the RQ's logical structure (distinct from its theoretical ambition):
| Type | Logic | Example |
| Existence | Does phenomenon X exist? | Can neonates perceive pain? |
| Description/Classification | What are the properties/types of X? | What types of identity configurations do hybrid entrepreneurs display? |
| Composition | What are the components of X? | What constitutes hybrid entrepreneurial identity work? |
| Relationship | How are X and Y associated? | How does role salience relate to venture commitment? |
| Comparative | How does X differ across conditions? | Are hybrid entrepreneurs' identity narratives more fluid than full-time entrepreneurs'? |
| Causal | Does X cause Y? | Does identity conflict reduce hybrid entrepreneurs' performance? |
| Causal-Comparative | Does X cause Y more than Z does? | Does identity integration improve outcomes more than identity separation? |
| Causal-Comparative-Interaction | Does X cause Y more than Z under condition C? | Does identity integration improve outcomes more than separation — but only under institutional ambiguity? |
Why this matters: RQ type determines study design, method, and the kind of evidence that can answer it. A causal RQ investigated with a cross-sectional survey is structurally misaligned. A relationship RQ treated as causal overstates findings. Name the type; flag misalignments with intended methodology.
Escalation logic: Existence → Description → Relationship → Causal is the natural progression of a research programme. Jumping to causal claims without relational grounding is a common overreach.
A3 — Assumptions Audit
Surface the hidden assumptions the current RQ leaves untouched. Ask:
- What must be true for this RQ to make sense as formulated?
- What does the existing literature take for granted that this RQ never questions?
- What would a hostile reviewer call "the unexamined premise"?
- What would disappear if the foundational construct were defined differently?
- Does the RQ presuppose that the theory's core mechanism operates — rather than asking whether it operates?
Write out at least 3, ideally 5. Name each assumption as a proposition, not a theme.
A4 — Tension Identification
For each assumption: what theoretical tradition, empirical finding, or logical argument contradicts it?
Flag the most productive tensions — those that are:
- Substantive, not merely definitional
- Resolvable through empirical or theoretical work
- Surprising enough that a reader's worldview shifts if the assumption fails
A5 — Upgraded RQ
Generate 2–3 alternative RQs. For each:
- State the assumption being challenged
- Explain why it matters if the assumption fails
- Formulate the RQ
- Label its spectrum position (track-bound ↔ disruptive)
- Label its RQ type (from A2)
- Note what theoretical and methodological moves it demands
The If-Not Question: The strongest move is to formulate the RQ so that both possible answers are theoretically significant. Instead of "How do X do Y?" ask "Do X do Y as theory predicts — and if not, what does that reveal about the theory's boundary conditions?" This treats the theory as hypothesis, not frame.
The Third-Category Move: When a field operates with two exhaustive categories, ask whether the phenomenon under study requires a third category that the binary forecloses. If so, name the candidate concept and build the RQ around it.
Mode B: Build a Research Question from Scratch
B1 — Map the Existing Assumption Landscape
Externalise as explicit propositions:
- What does a textbook in this field state as settled?
- What sampling decisions does the field treat as obvious?
- What outcome variables does the field treat as the right ones to measure?
- What temporal horizon does the field treat as the relevant one?
Each of these can hide a scope condition.
B2 — Run Assumption Stress Tests
For each consensus proposition:
- Reversal — What if the opposite were true?
- Scope challenge — Under what conditions does this assumption break down?
- Unit of analysis shift — What changes if we zoom in / zoom out / change the actor?
- Temporal challenge — Was this assumption always true? Is it becoming less true?
- Cross-paradigm intrusion — What would a researcher from an adjacent field say is missing?
- Population bias — Has this assumption only been tested on populations where it is structurally guaranteed to hold?
B3 — Identify the Most Productive Crack
Select the assumption whose challenge:
- Has empirical or theoretical warrant
- Has not been made before in this form
- Generates novel predictions or reframings
- Connects to observable phenomena
B4 — RQ Formulation
Structure as:
"Existing literature assumes [X]. But [evidence/argument] suggests [Y]. This raises the question: [RQ]."
This makes the problematization logic explicit and shows reviewers where the contribution lives.
Then: assign an RQ type (from A2) and proceed to Mode G (FINERMAPS Check) to verify executability.
Mode C: From Signal to RQ
C1 — Signal Crystallization
Push past surface description toward a one-sentence statement of the puzzle in terms that identify a theoretical tension, not just an empirical curiosity.
Example:
- Surface: "Hybrid entrepreneurs seem to behave differently than the literature predicts"
- Crystallized: "Role identity theory assumes salience hierarchies stabilise over time, but hybrid entrepreneurs appear to maintain fluid, context-dependent hierarchies indefinitely — suggesting stabilisation presupposes career resolution as a structural precondition the theory never made explicit"
C2 — Literature Anchoring
Which assumption in which literature does this puzzle put under pressure? Name it as a proposition.
C3 → Mode B (B3 → B4)
The crystallized signal is the productive crack. Proceed to RQ formulation.
Mode D: Review RQ
For literature review chapters that must carry an independent contribution.
A review RQ is not a summary question ("What do we know about X?"). It is a diagnostic question: the review itself becomes the finding. The literature's patterns — what it studied, how, with what populations, measuring which outcomes — constitute evidence for a claim about the field's implicit scope conditions.
D1 — Identify the Field's Implicit Scope Conditions
Ask:
- What populations has the field systematically studied? What populations are structurally absent?
- What mechanisms have been tested under what conditions? Could those conditions guarantee the mechanism works by design?
- What outcome variables does the field treat as obvious? Are these empirically validated criteria or normative assumptions?
- What temporal logics do study designs presuppose?
D2 — Formulate the Review RQ
Structure as a diagnostic claim, not a mapping exercise:
"What do [theoretical frameworks / study populations / outcome measures / methodological designs] in [field] collectively reveal about [the field's implicit assumptions] — and which [phenomena / configurations / mechanisms] do these assumptions render systematically invisible?"
D3 — Derive Subquestions
Each subquestion operationalises one dimension of the diagnostic claim across the review corpus:
- SQ-Population: Which configurations appear as primary populations vs. transitional baselines?
- SQ-Mechanism: Which mechanisms have been tested, and do their operationalisations presuppose the very conditions under investigation?
- SQ-Outcome: How is adequacy/success defined — empirically validated or normatively assumed?
- SQ-Temporal: What temporal horizon do study designs presuppose?
- SQ-Method: Are there systematic mismatches between dominant methods and the empirical demands of the phenomenon?
Each SQ should be answerable through systematic coding, not narrative synthesis alone.
Mode E: RQ Suite
For reformulating a full set of interdependent research questions so they form a coherent escalating problematization logic.
E1 — Audit the Suite as a Whole
Ask:
- Do the RQs collectively challenge the framework, or operate within it?
- Is there an escalating logic — each RQ going deeper than the previous?
- Does any RQ treat the theory's core mechanism as a hypothesis to be tested, or only as a lens to be applied?
- Does the overarching research problem carry the theoretical provocation, or only describe the empirical territory?
E2 — Assign Roles
A strong suite typically has:
- RQ1 — Tests whether the theory's core mechanism operates as predicted (If-Not Question)
- RQ2 — Examines what context, infrastructure, or condition shapes the phenomenon the theory cannot explain
- RQ3 — Challenges whether the field's outcome categories are adequate — opens the Third-Category move
E3 — Reformulate for Stringency
Each RQ must pass four tests:
- Falsifiability — Does the RQ carry a falsifiable theoretical claim, not just a descriptive intent?
- Stake — Is the assumption on the line named explicitly in the question?
- Bidirectional surprise — Would both possible answers be theoretically significant?
- Non-redundancy — Does each RQ attack a different layer of the same foundational assumption?
E4 — Check Internal Consistency
Verify that the overarching research problem, the review RQ (if present), and the empirical RQs share a single problematization logic. If the overarching problem labels an assumption but the empirical RQs operationalise it unchallenged, flag the contradiction explicitly.
Mode F: Stepwise Build from Topic
For users who have a broad area of interest but no RQ yet. Walks through development from topic → question → hypothesis.
F1 — Identify and Scope the Subject
Ask: What is the phenomenon? What population? What time horizon? What context? Narrow until the topic can be expressed as a single coherent sentence.
Common failure mode: the topic is too broad to have a question (not a gap in knowledge but an entire research programme).
F2 — Map Information Gaps
Three questions:
- What do you already know? (existing literature, established findings)
- What do you still need to know? (the gap)
- What are the implied questions — what must be answered first before the main question is even answerable?
Work from the most fundamental implied question upward.
F3 — Assess Question Quality: Bad vs. Good Patterns
Before proceeding, diagnose whether the forming question falls into common failure patterns:
| Failure Pattern | Diagnosis | Fix |
| Too narrow | Can be answered with a single statistic; answerable "yes/no" | Introduce a moderating variable, a comparative element, or a "how/why" framing |
| Too broad | Methodology would be impossible; not coverable in a single study | Specify population, time frame, mechanism, outcome variable |
| Too objective | Produces only descriptive factual data with no arguable position | Introduce a relationship, comparison, or causal claim |
| Too simple | Answerable via web search or existing databases without original data | Require investigation AND evaluation, not just retrieval |
| Unfocused | Multiple phenomena treated simultaneously without clear priority | One core concept, maximum two or three supporting concepts |
Upgrade moves:
- Too narrow → "How does [variable] affect [outcome] in [population]?" instead of "What is [statistic]?"
- Too broad → Specify one mechanism, one population, one temporal window
- Too simple → Require both data collection AND argument construction
F4 — Frame the RQ
Write the question as completely as possible. Then decompose:
- Core concept (max 2–3)
- Specified population
- Intervention or exposure, if any
- Outcome of interest
Then assign an RQ type (from A2). The type determines what kind of study can answer it.
F5 — Evaluate Against FINERMAPS
Proceed to Mode G.
F6 — Upgrade to Problematization
If the RQ passes FINERMAPS but remains gap-spotting, proceed to Mode B (B1–B4) to push toward genuine problematization.
Mode G: FINERMAPS Quality Check
Applied to any finalised or near-finalised RQ as an operational quality gate. A theoretically sharp RQ still fails if it cannot be executed, approved, or published.
| Criterion | Check question | Common failure |
| Feasible | Can this be done with available subjects, methods, time, and resources? | Scope exceeds available access, sample size, or skills |
| Interesting | Does the researcher have genuine, grounded investment in this question? | Topic chosen for strategic reasons without intellectual commitment |
| Novel | Does this go beyond replication? Does it confirm, refute, extend, or reframe existing findings? | Copies prior RQ without differentiation; not clearly distinct from existing literature |
| Ethical | Can IRB/ethics clearance be obtained? Does it minimise harm, protect privacy, allow withdrawal? | Participant risk not addressed; deceptive elements present |
| Relevant | Does it arise from current literature, practice, or policy? Does it fill a gap or challenge an assumption that matters? | Interesting but disconnected from any field's active questions |
| Manageable | Can the researcher execute this independently within realistic constraints? | Requires collaboration or infrastructure the researcher doesn't have |
| Appropriate | Is it logically and scientifically appropriate for the institutional and community context? | Mismatched with available data, population, or disciplinary norms |
| Potential value | Does it have health/social/policy/theoretical impact? Is it publishable? Does it resolve a dilemma for at least one stakeholder group? | Contribution is incremental without clear application; too niche to publish |
| Systematic | Can it be answered through a structured, rule-governed research process while still allowing creative thinking? | Too open-ended for systematic inquiry; no clear method path |
Output of Mode G: A FINERMAPS scorecard with pass/fail/revise for each criterion, plus specific revision suggestions for any failed criterion.
Important: FINERMAPS failure does not negate theoretical contribution. A genuinely problematizing RQ that fails "Feasible" needs scope adjustment, not abandonment. Adjust the scope, not the intellectual ambition.
Mode H: Hypothesis Bridge
Converts a finalised RQ into operational hypotheses.
A hypothesis is a falsifiable predictive statement: it specifies the expected direction, nature, and population of the relationship being tested. It operationalises the RQ as a testable proposition, not a question.
H1 — Determine Hypothesis Type
| RQ type | Appropriate hypothesis form |
| Existence | Null hypothesis + alternative: "X does / does not exist in population Y" |
| Relationship | Directional: "X is positively/negatively associated with Y in population Z" |
| Comparative | "Group A exhibits significantly more/less X than Group B" |
| Causal | "X causes Y; specifically, [mechanism] mediates the relationship" |
| Causal-Comparative | "The effect of X on Y is stronger/weaker than the effect of Z on Y" |
H2 — Strong Hypothesis Criteria
A strong hypothesis:
- Gives insight into the RQ without fully pre-empting it
- Is testable and measurable by the proposed method
- Has a logical basis (theoretical derivation, prior empirical evidence, or logical necessity)
- Follows the most likely outcome — not the exceptional or preferred outcome
- Specifies population, direction, and if causal, mechanism
H3 — Null Hypothesis Formulation
For quantitative studies: generate the null hypothesis (no effect, no difference, no relationship) and the alternative. This forces the researcher to state exactly what it would mean to be wrong.
H4 — Link Back to Theory
Show how rejecting or failing to reject the null hypothesis speaks to the theoretical assumption identified in the RQ. A hypothesis disconnected from its theoretical stake is operationally correct but intellectually empty.
Stringency Principles for Final Formulation
Applied to every RQ before delivery:
- Replace "How do X do Y?" with "Do X do Y as theory predicts — and if not, what does this reveal about [specific assumption]?"
- Replace "Under what conditions does X relate to Y?" with a question that names the assumption the moderator logic presupposes
- Replace "What is the role of X in Y?" with a question that specifies which competing theoretical predictions are at stake
- Remove hedge phrases that dilute the theoretical stake: "explore", "examine", "investigate", "shed light on"
- Ensure the phenomenon appears in the RQ as something that tests the theory, not something the theory explains
Anti-Patterns: Full Register
| Anti-pattern | Symptom | Fix |
| Gap-as-problematization | "Little research has examined X" | What assumption made researchers not look at X? Challenge that assumption. |
| Critical vocabulary as cover | Foucault/poststructuralism applied to a neglect gap | The framework is the gap-fill, not the problematization |
| Quasi-problematization | "We challenge X by applying Y theory" | If Y was in your toolkit before the analysis, it's application spotting |
| Incremental extension | "We extend theory X to context Y" | Extension assumes X is correct. What if it isn't? |
| Assumption-labelling without assumption-challenging | Paper identifies foundational assumption as problematic, then uses it as primary analytical lens | The RQs must attack the assumption, not apply the theory built on it |
| Moderator-as-problematization | "Under what conditions does X relate to Y?" | Moderator questions accept the X–Y relationship as given. Challenge the relationship itself. |
| Incommensurability without stakes | "Existing literature is fundamentally flawed" | Too broad. Which specific assumption? What follows empirically if you're right? |
| Population-sampling-as-context | Studying a new population and calling it a new context | If the theory was only tested where its preconditions are guaranteed, the population gap IS the theoretical gap |
| Descriptive overarching problem | "How do X do Y and Z?" | The overarching research problem should carry the theoretical provocation, not describe the empirical territory |
| Too narrow operationally | RQ answerable with a single statistic | Introduce moderating variable, comparison, or mechanism |
| Too broad operationally | Impossible to cover in one study | Specify population, time frame, one mechanism |
| Type mismatch | Causal RQ with cross-sectional survey | Align RQ type with method capacity; downgrade claim or upgrade design |
Quality Criteria
Before finalising any RQ:
- Does the RQ challenge at least one assumption — not just apply theory to a new context?
- Would a defender of the existing literature feel genuinely challenged, not just asked to do more work?
- Is the assumption foundational enough that the field changes if it fails?
- Does the RQ generate genuine surprise if answered in either direction?
- Is the theoretical contribution distinct from the empirical contribution?
- Is it free of assumption-labelling-without-challenging?
- If it's a suite: does each RQ attack a different layer, escalating in depth?
- Has it passed FINERMAPS? (Mode G)
- Is the RQ type correctly identified and aligned with intended methodology?
- If a hypothesis is needed: does it specify population, direction, and theoretical stake?
Output Format
Always conclude with:
Diagnosis — named Sandberg/Alvesson category, RQ type, quoted rhetoric, identified anti-patterns Assumptions Audit — 3–5 explicit propositions left unchallenged The productive tension — 1–2 sentences, the intellectual core Candidate RQs — 2–3, with spectrum position, RQ type, and assumption-at-stake labelled Recommended RQ with brief justification FINERMAPS check — pass/revise per criterion (abbreviated if all pass) What this demands — theoretically and methodologically Hypothesis (if requested or Mode H triggered) — null + alternative + theoretical stake
For Mode E (suite): present the full reformulated suite with the escalating logic made explicit. For Mode F (stepwise): walk through each step before presenting the final RQ.
Sprache folgt dem Input (Deutsch / English).
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

