From compound-science
Explore methodological approaches through structured analysis before planning implementation
npx claudepluginhub james-traina/science-plugins --plugin compound-scienceThis skill is limited to using the following tools:
**Pipeline mode:** This command operates fully autonomously. All decisions are made automatically.
Provides UI/UX resources: 50+ styles, color palettes, font pairings, guidelines, charts for web/mobile across React, Next.js, Vue, Svelte, Tailwind, React Native, Flutter. Aids planning, building, reviewing interfaces.
Fetches up-to-date documentation from Context7 for libraries and frameworks like React, Next.js, Prisma. Use for setup questions, API references, and code examples.
Guides Payload CMS config (payload.config.ts), collections, fields, hooks, access control, APIs. Debugs validation errors, security, relationships, queries, transactions, hook behavior.
Pipeline mode: This command operates fully autonomously. All decisions are made automatically.
Brainstorming helps answer WHAT approach to take through structured analysis. It precedes /workflows:plan, which answers HOW to implement it.
Process knowledge: See references/brainstorming-techniques.md for detailed question techniques, approach exploration patterns, and parsimony principles.
<feature_description> #$ARGUMENTS </feature_description>
If the research question above is empty: Infer the question from recent context — open files, recent conversation, or the project's estimation code. If no context is available, state "No research question provided" and stop.
Evaluate whether brainstorming is needed based on the research question.
Clear requirements indicators:
If requirements are already clear:
Skip brainstorming and note: "Requirements are detailed enough to proceed directly to planning. Run /workflows:plan to continue." Then stop.
If requirements need exploration: Proceed to Phase 1.
Run a targeted scan to understand existing patterns and related methods:
Focus on: existing estimation code, identification strategies used in this project, methodology documented in papers or notes.
Analyze the research question systematically without user interaction:
Document findings from the research agent and decomposition. If the question is ambiguous, pick the most natural interpretation given the project context and note the assumption.
Propose 2-3 concrete methodological approaches based on research and analysis.
For each approach, provide:
| Criterion | Approach A | Approach B | Approach C |
|---|---|---|---|
| Description | 2-3 sentence summary | 2-3 sentence summary | 2-3 sentence summary |
| Theoretical properties | Consistency, efficiency, robustness to misspecification | ... | ... |
| Identification requirements | What assumptions are needed? How testable are they? | ... | ... |
| Computational cost | Estimation time, convergence difficulty, parallelizability | ... | ... |
| Data requirements | Sample size needs, variable availability, panel structure | ... | ... |
| Software availability | Packages (Python/R/Julia), maturity, documentation | ... | ... |
| Monte Carlo evidence | Finite-sample performance from methodology literature | ... | ... |
Recommendation: Select the simplest approach that satisfies the identification requirements. Apply parsimony — prefer well-understood methods with established software implementations over novel approaches unless the research question specifically demands novelty.
Document why the recommended approach was chosen and what conditions would favor the alternatives.
Entry condition: Phase 2 comparison table is complete with all criteria filled for all approaches. Exit condition: Document written to docs/brainstorms/ with all required YAML frontmatter fields.
Write the brainstorm output to docs/brainstorms/<topic>-requirements.md with YAML frontmatter:
---
status: active
date: YYYY-MM-DD
topic: <descriptive topic>
---
If docs/brainstorms/ contains a recent document matching this topic, ask the user: "Found existing brainstorm on this topic. Continue from it, or start fresh?"
Write a brainstorm document to docs/brainstorms/YYYY-MM-DD-<topic>-brainstorm.md.
Ensure docs/brainstorms/ directory exists before writing.
Document structure:
---
title: [Brainstorm Topic]
date: YYYY-MM-DD
status: complete
recommended-approach: [Name of recommended approach]
---
# [Brainstorm Topic]
## Research Question
[The question being explored]
## Problem Decomposition
- Core question: [...]
- Key constraints: [...]
- Prior art in this project: [...]
- Success criteria: [...]
## Approaches Compared
### Approach A: [Name]
- **Description:** [...]
- **Theoretical properties:** [...]
- **Identification requirements:** [...]
- **Computational cost:** [...]
- **Data requirements:** [...]
- **Software:** [...]
- **Monte Carlo evidence:** [...]
- **Verdict:** [...]
### Approach B: [Name]
[Same structure]
### Approach C: [Name] (if applicable)
[Same structure]
## Recommendation
**Selected: [Approach Name]**
[Why this approach. What conditions would favor alternatives. Key tradeoffs accepted.]
## Key Decisions
- [Decision 1 and rationale]
- [Decision 2 and rationale]
## Assumptions Made
- [Any assumptions made during autonomous analysis]
## Open Questions
- [Questions that should be resolved during planning or implementation]
## References
- [Methodological papers cited]
- [Software documentation referenced]
Pipeline mode (invoked from /lfg or /slfg):
/workflows:plan with the brainstorm document path as contextStandalone mode (invoked directly by the user):
Display the closing summary:
Brainstorm complete!
Document: docs/brainstorms/YYYY-MM-DD-<topic>-brainstorm.md
Recommended approach: [Approach Name]
Key rationale: [One-line summary]
Alternatives documented:
- [Alternative 1]: [When to prefer]
- [Alternative 2]: [When to prefer]
Then present these options:
What would you like to do next?
1. Proceed to planning (Recommended) — Run /workflows:plan now with this brainstorm
2. Refine the brainstorm — Revisit the Phase 2 comparison with updated criteria
3. End session — The brainstorm document is saved; pick up later with /workflows:plan
Wait for the user to choose:
/workflows:plan with the brainstorm document path as context./workflows:plan — proceed with the chosen approachliterature-scout agent — for deeper literature search on a specific methodmethods-explorer agent — for deep dives into estimator properties and software implementationsNEVER CODE! Just explore and document methodological decisions.