| name | idea-engine |
| description | Activate when the user needs to evaluate whether a research idea is worth pursuing, brainstorm new research directions, or stress-test a paper concept before committing. This is Phase 0 of the paper machine — the gate that decides whether the full 8-phase pipeline should run. Also activates standalone via /evaluate-idea.
Integrates Carlini's research philosophy (conclusion-first test, taste for problems, kill conditions, unreasonable effort) with structured evaluation (7 dimensions, RS1-RS8 principles, 3 specialist agents).
|
Idea Engine — Phase 0: Is This Paper Worth Writing?
Purpose
Most AI tools help you write papers. This engine helps you decide which papers to write.
The current paper machine takes any topic and produces. This engine adds the missing
gate: Phase 0 — Is this paper the right use of my time?
Great research starts with taste, strategic problem selection, honest self-evaluation,
and knowing when to kill your darlings.
When to Activate
- User runs
/evaluate-idea [topic]
- User asks "is this worth pursuing?", "should I write this paper?"
- Phase 0 of the
/write-paper pipeline (before Reconnaissance)
- User wants to compare multiple research directions
- User questions whether to continue an ongoing project
The 8 Research Strategy Principles (RS1-RS8)
Reference: principles/research-strategy.md for full details.
Problem Selection
- RS1 (Novelty Test): If you don't do this, how many months until someone else does?
- RS2 (Conclusion-First Test): Can you write a compelling conclusion without doing the work?
- RS3 (Nugget Test): Can you state the key insight in one sentence?
Execution Strategy
- RS4 (Fail Fast): Start with the sub-problem most likely to kill the project.
- RS5 (Kill Early): A working project with low impact is worse than a killed project.
- RS6 (Unreasonable Effort): Strengthen "sometimes" to "usually" — but only AFTER RS4 and RS5.
Strategic Positioning
- RS7 (Comparative Advantage): Research space is high-dimensional; find your unique corner.
- RS8 (Timing Awareness): Impact = skill x domain importance at this moment.
The 6-Phase Orchestration Pipeline
Phase 1: SEED — Understand the Problem Space
Goal: Gather enough context to brainstorm and evaluate intelligently.
- Check for prior evaluations in
research-evaluations/*.md
- If KILL exists: check whether reasons still hold
- If PARK exists: check whether revisit conditions are met
- Brief interview (3-5 questions max, skip what's already answered):
- What's the problem space? What's bugging you about it?
- What's your background/expertise relevant to this?
- Any constraints (timeline, venue, data availability)?
- What would success look like?
Phase 2: DIVERGE — Generate Ideas
Goal: Expand the possibility space before narrowing.
- Deploy the Brainstormer agent (
agents/brainstormer.md)
- Organize results by type:
- Cross-field connections
- Assumptions worth challenging
- Novel framings
- Extensions of the original idea
- Present results. User stars their top 2-3 ideas.
Phase 3: EVALUATE — Stress-Test Top Ideas
Goal: Honest, structured assessment of each candidate.
- Deploy Idea Critic agents in parallel — one per selected idea (
agents/idea-critic.md)
- Each evaluation covers all 7 dimensions with signal ratings
- Present side-by-side comparison table:
| Dimension | Idea A | Idea B | Idea C |
|-------------|---------------|---------------|---------------|
| Novelty | Years | Months | Weeks |
| Impact | High | Medium | Low |
| Timing | Well-Timed | Well-Timed | Too Late |
| Feasibility | Medium Risk | Low Risk | Low Risk |
| Competition | Open | Moderate | Crowded |
| Nugget | Clear | Fuzzy | Clear |
| Narrative | Compelling | Workable | Weak |
| **VERDICT** | **PURSUE** | **REFINE** | **KILL** |
- Highlight which ideas survived and which were killed, with reasoning.
Phase 4: DEEPEN — Strategic Reality Check
Goal: For surviving ideas, assess strategic viability.
- Deploy Research Strategist for each PURSUE/REFINE idea (in parallel,
agents/research-strategist.md)
- Scooping risk assessment
- Competitive landscape analysis
- Timing evaluation
- Comparative advantage mapping
- Present as reality check with traffic lights:
- Green flags (strong strategic position)
- Yellow flags (manageable risks)
- Red flags (serious concerns)
Phase 5: FRAME — The Conclusion-First Test
Goal: The decisive test. Can this paper tell a compelling story?
For each surviving idea, write:
- The Nugget — one sentence, the core insight
- Draft Abstract (5 sentences following Carlini's Formula A):
- (1) Topic
- (2) Problem within that topic
- (3) Results or methods
- (4) Whichever sentence 3 didn't cover
- (5) Why it matters
- Draft Conclusion (2-3 sentences):
- What can we say? What changed? So what?
The critical test: If the conclusion feels hollow or generic — if it only says
"our method achieves X% improvement" — that IS the signal. The idea doesn't have
enough impact to justify months of work.
Phase 6: DECIDE — Final Verdict and Next Steps
Goal: Clear, actionable decision for each idea.
For each idea, deliver one of three verdicts:
PURSUE — This is worth your time. Next steps:
- Specific, risk-targeted first step (RS4: start with what's most likely to kill it)
- Time-bounded: achievable in 1-2 weeks
- What to do first, what to defer
PARK — Not now, but potentially later. Document:
- What would need to change for this to become PURSUE
- Conditions to revisit (new dataset, field shift, new collaborator)
- Set a revisit date
KILL — Not worth pursuing. But still extract value:
- What was learned from the evaluation process
- Any salvageable sub-ideas worth noting
- Blog post / workshop paper / conversation starter potential
Save evaluation to research-evaluations/YYYY-MM-DD-<topic-slug>.md:
---
date: YYYY-MM-DD
topic: [descriptive title]
verdict: [PURSUE/PARK/KILL]
nugget: [one-sentence key insight]
revisit: [date or condition, if PARK]
---
[table from Phase 3]
[top 3 risks]
[summary from Phase 4]
[from Phase 5]
[from verdict]
Orchestration Rules
- Maximize parallelism in Phases 3-4. Run multiple Idea Critic / Research Strategist
agents simultaneously for different ideas.
- Show your plan before each phase. Brief one-liner, not a detailed outline.
- Let the researcher drive. Present options, don't decide for them which ideas to pursue.
- Don't skip Phase 5 (conclusion-first test). This is the most important phase.
Everything else is preparation for this moment.
- Be honest in synthesis. If all ideas score poorly, say so. Don't manufacture enthusiasm.
- Keep momentum. A full evaluation session should take 15-20 minutes, not an hour.
If the user provides a single clear idea, collapse Phases 1-2 and go straight to evaluation.
- Connect to the pipeline. If verdict is PURSUE and user wants to proceed, hand off
directly to Phase 1 (Reconnaissance) of the paper machine. The evaluation artifacts
inform the literature search strategy.
Standalone vs. Pipeline Mode
Standalone (/evaluate-idea): Run the full 6-phase evaluation. End with verdict.
The user decides whether to proceed to /write-paper.
Pipeline (Phase 0 of /write-paper): Run an abbreviated evaluation:
- Skip Phase 2 (DIVERGE) if user already has a specific topic
- Run Phase 3 (EVALUATE) with the single idea
- Run Phase 5 (FRAME: conclusion-first test)
- If PURSUE: proceed to Phase 1 (Reconnaissance) automatically
- If KILL: stop and explain why. Suggest alternatives or refinements.
- If REFINE: present refinement suggestions, let user choose, then re-evaluate or proceed.
Integration with Paper Machine
When Phase 0 produces a PURSUE verdict, the following artifacts carry forward:
- The nugget → becomes the guiding star for the entire paper
- Draft conclusion → informs the framing in Phase 2
- Competitive landscape → focuses the literature search in Phase 1
- Key risks → determines what the paper must address to be convincing
- Draft abstract → first draft for Phase 4
