Idea Discovery Agent Skill

Workflow 1: Idea Discovery Pipeline

Orchestrate a complete idea discovery workflow for: $ARGUMENTS

Overview

This skill chains sub-skills into a single automated pipeline:

/research-lit → /idea-creator → /novelty-check → /research-review → /research-refine-pipeline
  (survey)      (brainstorm)    (verify novel)    (critical feedback)  (refine method + plan experiments)

Each phase builds on the previous one's output. The final deliverables are a validated idea-stage/IDEA_REPORT.md with ranked ideas, plus a refined proposal (refine-logs/FINAL_PROPOSAL.md) and experiment plan (refine-logs/EXPERIMENT_PLAN.md) for the top idea.

Constants

• PILOT_MAX_HOURS = 2 — Skip any pilot experiment estimated to take > 2 hours per GPU. Flag as "needs manual pilot" in the report.
• PILOT_TIMEOUT_HOURS = 3 — Hard timeout: kill any running pilot that exceeds 3 hours. Collect partial results if available.
• MAX_PILOT_IDEAS = 3 — Run pilots for at most 3 top ideas in parallel. Additional ideas are validated on paper only.
• MAX_TOTAL_GPU_HOURS = 8 — Total GPU budget across all pilots. If exceeded, skip remaining pilots and note in report.
• AUTO_PROCEED = true — If user doesn't respond at a checkpoint, automatically proceed with the best option after presenting results. Set to false to always wait for explicit user confirmation.
• REVIEWER_MODEL = gpt-5.6-sol — Model used via a secondary Codex agent. Must be an OpenAI model (e.g., gpt-5.6-sol, o3, gpt-4o). Passed to sub-skills.
• ARXIV_DOWNLOAD = false — When true, /research-lit downloads the top relevant arXiv PDFs during Phase 1. When false (default), only fetches metadata. Passed through to /research-lit.
• COMPACT = false — When true, generate compact summary files for short-context sessions and downstream skills. Writes idea-stage/IDEA_CANDIDATES.md.
• OUTPUT_DIR = idea-stage/ — All idea-stage outputs go here. Create the directory if it doesn't exist.
• REF_PAPER = false — Reference paper to base ideas on. Accepts a local PDF path, arXiv URL, or paper URL. When set, summarize it first and use it as idea-generation context.
• RENDER_HTML = true — When true (default), auto-render idea-stage/IDEA_REPORT.md to HTML at workflow end via /render-html. Uses --no-review because the source already received novelty + same-family provisional review. Set false to skip.

> 💡 These are defaults. Override by telling the skill, e.g., /idea-discovery "topic" — ref paper: https://arxiv.org/abs/2406.04329 or /idea-discovery "topic" — compact: true.

Pipeline

Phase 0: Load Research Brief (if available)

Before starting any other phase, check for a detailed research brief in the project:

• Look for RESEARCH_BRIEF.md in the project root or a path passed in $ARGUMENTS.
• If found, read it and extract:
• problem statement and context
• constraints: compute, data, timeline, venue
• what the user already tried and what did not work
• domain knowledge and non-goals
• existing results, if any
• Use this as the primary context for all subsequent phases; it replaces the one-line prompt when more specific.
• If both RESEARCH_BRIEF.md and one-line $ARGUMENTS exist, merge them: the brief has priority for details, and the argument sets the direction.

If no brief exists, proceed normally with $ARGUMENTS as the research direction.

Recommended template:

# Research Brief

## Problem Statement
[What problem are we trying to solve?]

## Context
[Relevant field, current approach, why this matters]

## Constraints
- Compute:
- Data:
- Timeline:
- Target venue:

## What We Already Tried
- [attempt] -> [outcome]

## Non-Goals
- [what not to pursue]

Phase 0.5: Reference Paper Summary (when REF_PAPER is set)

Skip entirely if REF_PAPER is false.

Summarize the reference paper before searching the literature:

• If arXiv URL — invoke /arxiv "ARXIV_ID" — download to fetch the PDF, then read the first 5 pages.
• If local PDF path — read the PDF directly, focusing on the title, abstract, introduction, and method overview.
• If other URL — fetch the content and extract the method, results, and limitations.
• Generate idea-stage/REF_PAPER_SUMMARY.md using this template:

# Reference Paper Summary

## What They Did
[2-3 sentences: core method and contribution]

## Key Results
[Main quantitative findings]

## Limitations & Open Questions
[Acknowledged weaknesses, missing experiments, future work]

## Potential Improvement Directions
[Concrete ways to extend, challenge, or improve the paper]

## Codebase
[If `base repo` is set: link to the repo and identify relevant entry points]

Use idea-stage/REF_PAPER_SUMMARY.md as additional context in both Phase 1 and Phase 2.

Phase 1: Literature Survey

Invoke /research-lit to map the research landscape:

/research-lit "$ARGUMENTS" — composed: idea-stage/IDEA_REPORT.md

What this does:

• Search arXiv, Google Scholar, Semantic Scholar for recent papers
• Build a landscape map: sub-directions, approaches, open problems
• Identify structural gaps and recurring limitations
• Output a literature summary (saved to working notes)

🚦 Checkpoint: Present the landscape summary to the user. Ask:

📚 Literature survey complete. Here's what I found:
- [key findings, gaps, open problems]

Does this match your understanding? Should I adjust the scope before generating ideas?
(If no response, I'll proceed with the top-ranked direction.)

• User approves (or no response + AUTO_PROCEED=true) → proceed to Phase 2 with best direction.
• User requests changes (e.g., "focus more on X", "ignore Y", "too broad") → refine the search with updated queries, re-run /research-lit with adjusted scope, and present again. Repeat until the user is satisfied.

Phase 2: Idea Generation + Filtering + Pilots

Invoke /idea-creator with the landscape context and idea-stage/REF_PAPER_SUMMARY.md if available:

/idea-creator "$ARGUMENTS" — composed: idea-stage/IDEA_REPORT.md

What this does:

• If idea-stage/REF_PAPER_SUMMARY.md exists, include it as context so ideas explicitly build on, improve, or extend the reference paper
• Brainstorm 8-12 concrete ideas via GPT-5.6-Sol xhigh
• Filter by feasibility, compute cost, quick novelty search
• Deep validate top ideas (full novelty check + devil's advocate)
• Run parallel pilot experiments on available GPUs (top 2-3 ideas)
• Rank by empirical signal
• Output idea-stage/IDEA_REPORT.md

🚦 Checkpoint: Present idea-stage/IDEA_REPORT.md ranked ideas to the user. Ask:

💡 Generated X ideas, filtered to Y, piloted Z. Top results:

1. [Idea 1] — Pilot: POSITIVE (+X%)
2. [Idea 2] — Pilot: WEAK POSITIVE (+Y%)
3. [Idea 3] — Pilot: NEGATIVE, eliminated

Which ideas should I validate further? Or should I regenerate with different constraints?
(If no response, I'll proceed with the top-ranked ideas.)

• User picks ideas (or no response + AUTO_PROCEED=true) → proceed to Phase 3 with top-ranked ideas.
• User unhappy with all ideas → collect feedback ("what's missing?", "what direction do you prefer?"), update the prompt with user's constraints, and re-run Phase 2 (idea generation). Repeat until the user selects at least 1 idea.
• User wants to adjust scope → go back to Phase 1 with refined direction.

Phase 3: Deep Novelty Verification

For each top idea (positive pilot signal), run a thorough novelty check:

/novelty-check "[top idea 1 description]"
/novelty-check "[top idea 2 description]"

What this does:

• Multi-source literature search (arXiv, Scholar, Semantic Scholar)
• Cross-verify with GPT-5.6-Sol xhigh
• Check for concurrent work (last 3-6 months)
• Identify closest existing work and differentiation points

Update idea-stage/IDEA_REPORT.md with deep novelty results. Eliminate any idea that turns out to be already published.

Phase 4: External Critical Review

For the surviving top idea(s), get brutal feedback:

/research-review "[top idea with hypothesis + pilot results]" — composed: idea-stage/IDEA_REPORT.md

What this does:

• GPT-5.6-Sol xhigh acts as a senior reviewer (NeurIPS/ICML level)
• Scores the idea, identifies weaknesses, suggests minimum viable improvements
• Provides concrete feedback on experimental design

Update idea-stage/IDEA_REPORT.md with reviewer feedback and revised plan.

idea-stage/IDEA_REPORT.md is this pipeline's one canonical deliverable. The

explicit — composed: signal makes each sub-skill return/fold unique findings

instead of scattering LIT_LANDSCAPE.md, RESEARCH_REVIEW.md, or duplicate

manifests. Without that signal, every sub-skill remains standalone. See

output-composition.md.

Phase 4.5: Method Refinement + Experiment Planning

After review, refine the top idea into a concrete proposal and plan experiments:

/research-refine-pipeline "[top idea description + pilot results + reviewer feedback]"

What this does:

• Freeze a Problem Anchor to prevent scope drift
• Iteratively refine the method via GPT-5.6-Sol review (up to 5 rounds, until score ≥ 9)
• Generate a claim-driven experiment roadmap with ablations, budgets, and run order
• Output: refine-logs/FINAL_PROPOSAL.md, refine-logs/EXPERIMENT_PLAN.md, refine-logs/EXPERIMENT_TRACKER.md

🚦 Checkpoint: Present the refined proposal summary:

🔬 Method refined and experiment plan ready:
- Problem anchor: [anchored problem]
- Method thesis: [one sentence]
- Dominant contribution: [what's new]
- Must-run experiments: [N blocks]
- First 3 runs to launch: [list]

Proceed to implementation? Or adjust the proposal?

• User approves (or AUTO_PROCEED=true) → proceed to Final Report.
• User requests changes → pass feedback to /research-refine for another round.
• Lite mode: If reviewer score < 6 or pilot was weak, run /research-refine only (skip /experiment-plan) and note remaining risks in the report.

Phase 5: Final Report

Finalize idea-stage/IDEA_REPORT.md with all accumulated information:

# Idea Discovery Report

**Direction**: $ARGUMENTS
**Date**: [today]
**Pipeline**: research-lit → idea-creator → novelty-check → research-review → research-refine-pipeline

## Executive Summary
[2-3 sentences: best idea, key evidence, recommended next step]

## Literature Landscape
[from Phase 1]

## Ranked Ideas
[from Phase 2, updated with Phase 3-4 results]

### 🏆 Idea 1: [title] — RECOMMENDED
- Pilot: POSITIVE (+X%)
- Novelty: CONFIRMED (closest: [paper], differentiation: [what's different])
- Reviewer score: X/10
- Next step: implement full experiment → /auto-review-loop

### Idea 2: [title] — BACKUP
...

## Eliminated Ideas
[ideas killed at each phase, with reasons]

## Refined Proposal
- Proposal: `refine-logs/FINAL_PROPOSAL.md`
- Experiment plan: `refine-logs/EXPERIMENT_PLAN.md`
- Tracker: `refine-logs/EXPERIMENT_TRACKER.md`

## Next Steps
- [ ] /run-experiment to deploy experiments from the plan
- [ ] /auto-review-loop to iterate until submission-ready
- [ ] Or invoke /research-pipeline for the complete end-to-end flow

Phase 5.5: Write Compact Files (when COMPACT = true)

Skip entirely if COMPACT is false.

Write idea-stage/IDEA_CANDIDATES.md — a lean summary of the top 3-5 surviving ideas:

# Idea Candidates

| # | Idea | Pilot Signal | Novelty | Reviewer Score | Status |
|---|------|-------------|---------|---------------|--------|
| 1 | [title] | +X% | Confirmed | X/10 | RECOMMENDED |
| 2 | [title] | +Y% | Confirmed | X/10 | BACKUP |
| 3 | [title] | Negative | — | — | ELIMINATED |

## Active Idea: #1 — [title]
- Hypothesis: [one sentence]
- Key evidence: [pilot result]
- Next step: /experiment-bridge or /research-refine

Phase 5.6: Instantiate the Research Contract (always — NOT gated on COMPACT)

When Phase 4 ends with a RECOMMENDED idea, create idea-stage/docs/research_contract.md

from templates/RESEARCH_CONTRACT_TEMPLATE.md (repo root or $ARIS_REPO/templates/),

filling in: the selected idea + selection rationale, core claims, minimum

convincing evidence, and the next-step pointer. Skip only when the run produced

no RECOMMENDED idea. /experiment-bridge implements against this contract;

/result-to-claim + /ablation-planner read it as the claims source; session

recovery reloads the ACTIVE idea from it instead of the full idea pool.

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

Render HTML view (auto, when RENDER_HTML = true)

After finalizing idea-stage/IDEA_REPORT.md (and the optional IDEA_CANDIDATES.md), invoke /render-html on the report so the user has a single-file HTML view for tablet / phone reading:

/render-html "idea-stage/IDEA_REPORT.md" --no-review

--no-review is intentional: source MD already received this skill's novelty + same-family provisional review. HTML render is a structural conversion, not a new claim-audit gate.

Non-blocking: if /render-html fails (helper missing, secondary Codex agent unavailable, file write error), log the failure and continue. Skip entirely if RENDER_HTML = false.

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.

• Don't skip phases. Each phase filters and validates — skipping leads to wasted effort later.
• Checkpoint between phases. Briefly summarize what was found before moving on.
• Kill ideas early. It's better to kill 10 bad ideas in Phase 3 than to implement one and fail.
• Empirical signal > theoretical appeal. An idea with a positive pilot outranks a "sounds great" idea without evidence.
• Document everything. Dead ends are just as valuable as successes for future reference.
• Be honest with the reviewer. Include negative results and failed pilots in the review prompt.
• Feishu notifications are optional. If ~/.codex/feishu.json exists, send checkpoint at each phase transition and pipeline_done at final report. If absent/off, skip silently.

Composing with Workflow 2

After this pipeline produces a validated top idea:

/idea-discovery "direction"         ← you are here (Workflow 1, includes method refinement + experiment planning)
/run-experiment                     ← deploy experiments from the plan
/auto-review-loop "top idea"        ← Workflow 2: iterate until submission-ready

Or use /research-pipeline for the full end-to-end flow.