Novelty Check Agent Skill

> Override for Codex users who want Claude Code, not a second Codex agent, to act as the reviewer. Install this package after skills/skills-codex/*.

>

> This reviewer is a different model family from the Codex executor. Every overlay trace/audit records:

>

> `yaml

> review_independence: cross-family

> acceptance_status: accepted

> `

Novelty Check Skill

Check whether a proposed method/idea has already been done in the literature: $ARGUMENTS

Constants

• REVIEWER_MODEL = claude-review — Claude reviewer invoked through the local claude-review MCP bridge. Set CLAUDE_REVIEW_MODEL if you need a specific Claude model override.
• REVIEWER_BACKEND = claude-review — reviews route through the claude-review MCP (Claude family; cross-family for a Codex executor).

Instructions

Given a method description, systematically verify its novelty:

Phase A: Extract Key Claims

• Read the user's method description
• Identify 3-5 core technical claims that would need to be novel:
• What is the method?
• What problem does it solve?
• What is the mechanism?
• What makes it different from obvious baselines?

Phase B: Multi-Source Literature Search

For EACH core claim, search using ALL available sources:

• Web Search (via WebSearch):
• Search arXiv, Google Scholar, Semantic Scholar
• Use specific technical terms from the claim
• Try at least 3 different query formulations per claim
• Include year filters for 2024-2026

• Known paper databases: Check against:
• ICLR 2025/2026, NeurIPS 2025, ICML 2025/2026
• Recent arXiv preprints (2025-2026)

• Read abstracts: For each potentially overlapping paper, WebFetch its abstract and related work section

Phase C: Fresh-Agent Verification (cross-family accepted by default)

Call REVIEWER_MODEL via mcp__claude-review__review_start with high-rigor review:

mcp__claude-review__review_start:
  prompt: |
    [Full novelty briefing + prior work list + specific novelty questions]

After this start call, immediately save the returned jobId and poll mcp__claude-review__review_status with a bounded waitSeconds until done=true. Treat the completed status payload's response as the reviewer output, and save the completed threadId for any follow-up round.

Prompt should include:

• The proposed method description
• All papers found in Phase B
• Ask: "Is this method novel? What is the closest prior work? What is the delta?"

Phase D: Novelty Report

Output a structured report:

## Novelty Check Report

### Proposed Method
[1-2 sentence description]

### Core Claims
1. [Claim 1] — Novelty: HIGH/MEDIUM/LOW — Closest: [paper]
2. [Claim 2] — Novelty: HIGH/MEDIUM/LOW — Closest: [paper]
...

### Closest Prior Work
| Paper | Year | Venue | Overlap | Key Difference |
|-------|------|-------|---------|----------------|

### Overall Novelty Assessment
- Score: X/10
- Recommendation: PROCEED / PROCEED WITH CAUTION / ABANDON
- Key differentiator: [what makes this unique, if anything]
- Risk: [what a reviewer would cite as prior work]

### Suggested Positioning
[How to frame the contribution to maximize novelty perception]

Important Rules

• Be BRUTALLY honest — false novelty claims waste months of research time
• "Applying X to Y" is NOT novel unless the application reveals surprising insights
• Check both the method AND the experimental setting for novelty
• If the method is not novel but the FINDING would be, say so explicitly
• Always check the most recent 6 months of arXiv — the field moves fast

Review Tracing

After each mcp__claude-review__review_start or optional oracle-pro reviewer call, save the trace following ../shared-references/review-tracing.md. Write files directly to .aris/traces/novelty-check/<date>_run<NN>/ and record searched claims, closest papers, reviewer route, raw response, and final novelty decision. Respect the --- trace: parameter when present (default: full).