Aaai26 ai Review

    2

    A multi-stage, evidence-grounded academic review agent modeled on the AAAI-26 AI Review Pilot research.

    Free

    22 uses4.5 (2 reviews)

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    aaai26-ai-review

    Example session with this skill installed

    Review the attached PDF of my Latent SDE paper following the AAAI-26 pilot logic. Focus on the technical soundness of my theorems and empirical comparisons in Table 2.

    • Read your context and instructions
    • Compiled the aaai26-ai-review

    Strengths

    • Novel application of Latent SDEs to time-series forecasting.

    Weaknesses

    • Major: Missing comparison with [Source B] in Table 2.
    • Major: Theorem 3.1 assumes Lipschitz continuity without proof.
    • Minor: Notational conflict between Eq. 4 and Eq. 7 regarding the 'lambda' term.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Generate high-quality AI-assisted peer reviews modeled on the publicly described logic of the AAAI-26 AI Review Pilot. This skill implements a multi-stage review pipeline: it decomposes the paper into specialized analysis dimensions (contributions, methodology, experiments, presentation), synthesizes findings, self-critiques the draft for unsupported claims, and runs a final quality check for anonymity leaks and citation hallucinations.

    Output follows a structured format (Synopsis, Summary of Review, Strengths, Weaknesses with Major/Minor labels, References) — no accept/reject verdict by default, mirroring the AAAI-26 design philosophy of AI-as-assistance rather than AI-as-decision-maker.

    Compatible with PDF input. Works for any venue: AAAI, NeurIPS, ICML, ACL, CVPR, and more.

    • Cannot access external non-public databases for citation verification.
    • Accuracy depends on PDF OCR quality for complex formulas.
    • Does not provide final acceptance scores.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    4.5
    2 reviews
    5
    1
    4
    1
    3
    0
    2
    0
    1
    0

    This skill has been used 22 times.

    Trust & safety

    Security scan

    Security scan passed 7 months ago

    • Free with an Agensi account

    Permissions

    No special permissions declared or detected

    Listed7 months ago

    What's inside

    Frequently Asked Questions

    The agent uses a multi-stage pipeline that breaks papers down into specific dimensions (methodology, experiments, etc.) and performs internal self-critiques to ensure all claims are grounded in the paper's actual evidence, reducing common AI hallucinations.

    The skill is designed as a modular tool compatible with major LLM-based agent platforms that support PDF parsing, including standard frameworks used for academic research automation.

    No, following the AAAI-26 pilot philosophy, this agent provides detailed evaluative synthesis and structured feedback, but it does not issue a final 'Accept' or 'Reject' recommendation to ensure human-in-the-loop decision making.

    The output is formatted with a specific structure including a synopsis, summary, strengths, categorized weaknesses (Major/Minor), and a hallucination check for citations and anonymity leaks.

    While titled after the AAAI-26 pilot, the underlying logic is venue-agnostic and maintains the high standards required for top-tier conferences like NeurIPS, ICML, CVPR, and ACL.

    Your purchase includes the full multi-stage workflow configuration, periodic logic updates to align with evolving AI review research, and documentation on how to optimize PDF ingestion for the best review results.

    New in Code Quality & Review