Skip to content

idea-creator

Category: ideation
Field: —
License: MIT
Updated: 2026-05-18
Stages: rq-formulation · hypothesis-generation

Research Idea Creator

Generate publishable research ideas for: $ARGUMENTS

Overview

Given a broad research direction from the user, systematically generate, validate, and rank concrete research ideas. This skill composes with /research-lit, /novelty-check, and /research-review to form a complete idea discovery pipeline.

Constants

  • PILOT_MAX_HOURS = 2 — Skip any pilot estimated to take > 2 hours per GPU. Flag as "needs manual pilot".
  • PILOT_TIMEOUT_HOURS = 3 — Hard timeout: kill pilots exceeding 3 hours. Collect partial results if available.
  • MAX_PILOT_IDEAS = 3 — Pilot at most 3 ideas in parallel. Additional ideas are validated on paper only.
  • MAX_TOTAL_GPU_HOURS = 8 — Total GPU budget for all pilots combined.
  • REVIEWER_MODEL = gpt-5.5 — Model used via Codex MCP for brainstorming and review. Must be an OpenAI model (e.g., gpt-5.5, o3, gpt-4o).
  • REVIEWER_BACKEND = codex — Default: Codex MCP (xhigh). Override with — reviewer: oracle-pro for GPT-5.4 Pro via Oracle MCP. See shared-references/reviewer-routing.md.
  • OUTPUT_DIR = idea-stage/ — All idea-stage outputs go here. Create the directory if it doesn't exist.

💡 Override via argument, e.g., /idea-creator "topic" — pilot budget: 4h per idea, 20h total.

Workflow

Phase 0: Load Research Wiki (if active)

Skip this phase entirely if research-wiki/ does not exist.

If research-wiki/ exists, resolve the canonical helper using the shared resolution chain (see ../research-wiki/SKILL.md for the contract):

Bash
cd "$(git rev-parse --show-toplevel 2>/dev/null || pwd)" || exit 1
ARIS_REPO="${ARIS_REPO:-$(awk -F'\t' '$1=="repo_root"{print $2; exit}' .aris/installed-skills.txt 2>/dev/null)}"
WIKI_SCRIPT=".aris/tools/research_wiki.py"
[ -f "$WIKI_SCRIPT" ] || WIKI_SCRIPT="tools/research_wiki.py"
[ -f "$WIKI_SCRIPT" ] || { [ -n "${ARIS_REPO:-}" ] && WIKI_SCRIPT="$ARIS_REPO/tools/research_wiki.py"; }
[ -f "$WIKI_SCRIPT" ] || {
  echo "WARN: research_wiki.py not found at .aris/tools/, tools/, or \$ARIS_REPO/tools/." >&2
  echo "      The idea-creation primary output (idea ranking) will still be produced." >&2
  echo "      Wiki integration (load query_pack, write idea pages, add edges, rebuild query_pack) will be skipped." >&2
  echo "      Fix: rerun 'bash tools/install_aris.sh', export ARIS_REPO, or 'cp <ARIS-repo>/tools/research_wiki.py tools/'." >&2
  WIKI_SCRIPT=""
}
Text Only
if research-wiki/query_pack.md exists AND is less than 7 days old:
    Read query_pack.md and use it as initial landscape context:
    - Treat listed gaps as priority search seeds
    - Treat failed ideas as a banlist (do NOT regenerate similar ideas)
    - Treat top papers as known prior work (do not re-search them)
    Still run Phase 1 below for papers from the last 3-6 months (wiki may be stale)
else if research-wiki/ exists but query_pack.md is stale or missing:
    if [ -n "$WIKI_SCRIPT" ]: python3 "$WIKI_SCRIPT" rebuild_query_pack research-wiki/
    Then read query_pack.md as above

Phase 1: Landscape Survey (5-10 min)

Map the research area to understand what exists and where the gaps are.

  1. Scan local paper library first: Check papers/ and literature/ in the project directory for existing PDFs. Read first 3 pages of relevant papers to build a baseline understanding before searching online. This avoids re-discovering what the user already knows.

  2. Search recent literature using WebSearch:

  3. Top venues in the last 2 years (NeurIPS, ICML, ICLR, ACL, EMNLP, etc.)
  4. Recent arXiv preprints (last 6 months)
  5. Use 5+ different query formulations

  6. Read abstracts and introductions of the top 10-15 papers

  7. Build a landscape map:

  8. Group papers by sub-direction / approach
  9. Identify what has been tried and what hasn't
  10. Note recurring limitations mentioned in "Future Work" sections

  11. Flag any open problems explicitly stated by multiple papers

  12. Identify structural gaps:

  13. Methods that work in domain A but haven't been tried in domain B
  14. Contradictory findings between papers (opportunity for resolution)
  15. Assumptions that everyone makes but nobody has tested
  16. Scaling regimes that haven't been explored
  17. Diagnostic questions that nobody has asked

Phase 2: Idea Generation (brainstorm with external LLM)

Use the external LLM via Codex MCP for divergent thinking:

Text Only
mcp__codex__codex:
  model: REVIEWER_MODEL
  config: {"model_reasoning_effort": "xhigh"}
  prompt: |
    You are a senior ML researcher brainstorming research ideas.

    Research direction: [user's direction]

    Here is the current landscape:
    [paste landscape map from Phase 1]

    Key gaps identified:
    [paste gaps from Phase 1]

    Generate 8-12 concrete research ideas. For each idea:
    1. One-sentence summary
    2. Core hypothesis (what you expect to find and why)
    3. Minimum viable experiment (what's the cheapest way to test this?)
    4. Expected contribution type: empirical finding / new method / theoretical result / diagnostic
    5. Risk level: LOW (likely works) / MEDIUM (50-50) / HIGH (speculative)
    6. Estimated effort: days / weeks / months

    Prioritize ideas that are:
    - Testable with moderate compute (8x RTX 3090 or less)
    - Likely to produce a clear positive OR negative result (both are publishable)
    - Not "apply X to Y" unless the application reveals genuinely surprising insights
    - Differentiated from the 10-15 papers above

    Be creative but grounded. A great idea is one where the answer matters regardless of which way it goes.

Save the threadId for follow-up.

Phase 3: First-Pass Filtering

For each generated idea, quickly evaluate:

  1. Feasibility check: Can we actually run this experiment with available resources?
  2. Compute requirements (estimate GPU-hours)
  3. Data availability
  4. Implementation complexity

  5. Skip ideas requiring > 1 week of GPU time or unavailable datasets

  6. Novelty quick-check: For each idea, do 2-3 targeted searches to see if it's already been done. Full /novelty-check comes later for survivors.

  7. Impact estimation: Would a reviewer care about the result?

  8. "So what?" test: if the experiment succeeds, does it change how people think?
  9. Is the finding actionable or just interesting?

Eliminate ideas that fail any of these. Typically 8-12 ideas reduce to 4-6.

Phase 4: Deep Validation (for top ideas)

For each surviving idea, run a deeper evaluation:

  1. Novelty check: Use the /novelty-check workflow (multi-source search + GPT-5.4 cross-verification) for each idea

  2. Critical review: Use GPT-5.4 via mcp__codex__codex-reply (same thread):

    Text Only
    Here are our top ideas after filtering:
[paste surviving ideas with novelty check results]

For each, play devil's advocate:
- What's the strongest objection a reviewer would raise?
- What's the most likely failure mode?
- How would you rank these for a top venue submission?
- Which 2-3 would you actually work on?

  3. Combine rankings: Merge your assessment with GPT-5.4's ranking. Select top 2-3 ideas for pilot experiments.

Phase 5: Parallel Pilot Experiments (for top 2-3 ideas)

Before committing to a full research effort, run cheap pilot experiments to get empirical signal. This is the key differentiator from paper-only validation.

  1. Design pilots: For each top idea, define the minimal experiment that would give a positive or negative signal:
  2. Single seed, small scale (e.g., small dataset subset, fewer epochs)
  3. Target: 30 min - PILOT_MAX_HOURS per pilot on 1 GPU
  4. Estimate GPU-hours BEFORE launching. If estimated time > PILOT_MAX_HOURS, reduce scale (fewer epochs, smaller subset) or flag as "needs manual pilot"

  5. Clear success metric defined upfront (e.g., "if metric improves by > 1%, signal is positive")

  6. Deploy in parallel: Use /run-experiment to launch pilots on different GPUs simultaneously:

    Text Only
    GPU 0: Pilot for Idea 1
GPU 1: Pilot for Idea 2
GPU 2: Pilot for Idea 3
    Use run_in_background: true to launch all at once.

  7. Collect results: Use /monitor-experiment to check progress. If any pilot exceeds PILOT_TIMEOUT_HOURS, kill it and collect partial results. Once all pilots complete (or timeout), compare:

  8. Which ideas showed positive signal?
  9. Which showed null/negative results? (eliminate or deprioritize)
  10. Any surprising findings that suggest a pivot?

  11. Total GPU-hours consumed (track against MAX_TOTAL_GPU_HOURS budget)

  12. Re-rank based on empirical evidence: Update the idea ranking using pilot results. An idea with strong pilot signal jumps ahead of a theoretically appealing but untested idea.

Note: Skip this phase if the ideas are purely theoretical or if no GPU is available. Flag skipped ideas as "needs pilot validation" in the report.

Phase 6: Output — Ranked Idea Report

Write a structured report to idea-stage/IDEA_REPORT.md:

Markdown
## Research Idea Report

**Direction**: [user's research direction]
**Generated**: [date]
**Ideas evaluated**: X generated → Y survived filtering → Z piloted → W recommended

### Landscape Summary
[3-5 paragraphs on the current state of the field]

### Recommended Ideas (ranked)

#### Idea 1: [title]
- **Hypothesis**: [one sentence]
- **Minimum experiment**: [concrete description]
- **Expected outcome**: [what success/failure looks like]
- **Novelty**: X/10 — closest work: [paper]
- **Feasibility**: [compute, data, implementation estimates]
- **Risk**: LOW/MEDIUM/HIGH
- **Contribution type**: empirical / method / theory / diagnostic
- **Pilot result**: [POSITIVE: metric +X% / NEGATIVE: no signal / SKIPPED: needs GPU]
- **Reviewer's likely objection**: [strongest counterargument]
- **Why we should do this**: [1-2 sentences]

#### Idea 2: [title]
...

### Eliminated Ideas (for reference)
| Idea | Reason eliminated |
|------|-------------------|
| ... | Already done by [paper] |
| ... | Requires > 1 week GPU time |
| ... | Result wouldn't be interesting either way |

### Pilot Experiment Results
| Idea | GPU | Time | Key Metric | Signal |
|------|-----|------|------------|--------|
| Idea 1 | GPU 0 | 45 min | +2.3% CE | POSITIVE |
| Idea 2 | GPU 1 | 30 min | -0.1% CE | NEGATIVE |
| Idea 3 | GPU 2 | 1.5 hr | +0.8% CE | WEAK POSITIVE |

### Suggested Execution Order
1. Start with Idea 1 (positive pilot signal, lowest risk)
2. Idea 3 as backup (weak signal, may need larger scale to confirm)
3. Idea 2 eliminated by pilot — negative result documented

### Next Steps
- [ ] Scale up Idea 1 to full experiment (multi-seed, full dataset)
- [ ] If confirmed, invoke /auto-review-loop for full iteration

Phase 7: Write Ideas to Research Wiki (if active)

Skip this phase entirely if research-wiki/ does not exist.

This is critical for spiral learning — without it, ideas/ stays empty and re-ideation has no memory.

$WIKI_SCRIPT was resolved in Phase 0 above. If Phase 0 did not run (no research-wiki/), this phase is skipped. If Phase 0 ran but the resolution chain failed to find the helper ($WIKI_SCRIPT is empty), the page-write step still runs (idea pages are plain markdown the agent writes directly), but the edge / query-pack / log steps that require the helper are skipped with a single warning.

Text Only
if research-wiki/ exists:
    for each idea in recommended_ideas + eliminated_ideas:
        1. Create page: research-wiki/ideas/<idea_id>.md
           - node_id: idea:<id>
           - stage: proposed (or: piloted, archived)
           - outcome: unknown (or: negative, mixed, positive)
           - based_on: [paper:<slug>, ...]
           - target_gaps: [gap:<id>, ...]
           - Include: hypothesis, proposed method, expected outcome
           - If pilot was run: actual outcome, failure notes, reusable components

        2. Add edges (only if $WIKI_SCRIPT resolved):
           [ -n "$WIKI_SCRIPT" ] && python3 "$WIKI_SCRIPT" add_edge research-wiki/ --from "idea:<id>" --to "paper:<slug>" --type inspired_by --evidence "..."
           [ -n "$WIKI_SCRIPT" ] && python3 "$WIKI_SCRIPT" add_edge research-wiki/ --from "idea:<id>" --to "gap:<id>" --type addresses_gap --evidence "..."

    Rebuild query pack (only if $WIKI_SCRIPT resolved):
        [ -n "$WIKI_SCRIPT" ] && python3 "$WIKI_SCRIPT" rebuild_query_pack research-wiki/
    Log (only if $WIKI_SCRIPT resolved):
        [ -n "$WIKI_SCRIPT" ] && python3 "$WIKI_SCRIPT" log research-wiki/ "idea-creator wrote N ideas (M recommended, K eliminated)"

    if [ -z "$WIKI_SCRIPT" ]:
        echo "WARN: idea pages were written but edges / query_pack / log were skipped because research_wiki.py is unreachable (see Phase 0 warning above)." >&2

Output Protocols

Follow these shared protocols for all output files: - Output Versioning Protocol — write timestamped file first, then copy to fixed name - Output Manifest Protocol — log every output to MANIFEST.md - Output Language Protocol — respect the project's language setting

Key Rules

  • Large file handling: If the Write tool fails due to file size, immediately retry using Bash (cat << 'EOF' > file) to write in chunks. Do NOT ask the user for permission — just do it silently.

  • The user provides a DIRECTION, not an idea. Your job is to generate the ideas.

  • Quantity first, quality second: brainstorm broadly, then filter ruthlessly.
  • A good negative result is just as publishable as a positive one. Prioritize ideas where the answer matters regardless of direction.
  • Don't fall in love with any idea before validating it. Be willing to kill ideas.
  • Always estimate compute cost. An idea that needs 1000 GPU-hours is not actionable for most researchers.
  • "Apply X to Y" is the lowest form of research idea. Push for deeper questions.
  • Include eliminated ideas in the report — they save future time by documenting dead ends.
  • If the user's direction is too broad (e.g., "NLP", "computer vision", "reinforcement learning"), STOP and ask them to narrow it. A good direction is 1-2 sentences specifying the problem, domain, and constraint — e.g., "factorized gap in discrete diffusion LMs" or "sample efficiency of offline RL with image observations". Without sufficient specificity, generated ideas will be too vague to run experiments on.
  • Anti-hallucination for cited papers. When the landscape survey or novelty justification cites specific papers, every cited paper must pass pre-search verification (verify_papers.py, canonical name resolved per shared-references/integration-contract.md §2; 3-layer arXiv / CrossRef / S2 fallback inside the helper itself). Policy D1 (primary + degraded-output fallback): if the helper is unresolved or its invocation fails, mark candidates [UNVERIFIED] and continue rather than dropping or guessing. Never fabricate arXiv IDs, DOIs, or titles from memory. Full protocol in shared-references/citation-discipline.md § Pre-Search Verification Protocol.

Composing with Other Skills

After this skill produces the ranked report:

Text Only
/idea-creator "direction"     → ranked ideas
/novelty-check "top idea"     → deep novelty verification (already done in Phase 4, but user can re-run)
/research-review "top idea"   → external critical feedback
implement                     → write code
/run-experiment               → deploy to GPU
/auto-review-loop             → iterate until submission-ready

Review Tracing

After each mcp__codex__codex or mcp__codex__codex-reply reviewer call, save the trace following shared-references/review-tracing.md (Policy C — forensic; never silently skip). Use save_trace.sh (resolved per the chain in shared-references/integration-contract.md §2) or write files directly to .aris/traces/<skill>/<date>_run<NN>/. Respect the --- trace: parameter (default: full).