Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    papermate

    2

    A skeptical research partner for researchers that kills weak ideas, audits feasibility, and scaffolds experiments.

    Free

    4 uses3.0 (2 reviews)
    papermate

    papermate

    Example session with this skill installed

    I want to use LLMs for Bangla sentiment analysis on social media data. Is this a good thesis topic, and can I run it on a free Colab GPU?

    • Read your context and instructions
    • Compiled the papermate

    I have analyzed your idea for Bangla sentiment analysis using LLMs. While the task is high-impact, the "LLM + Sentiment" combination is heavily saturated. ## Idea Autopsy: Bangla Sentiment with LLMs | Dimension | Score | Justification | | :--- | :--- | :--- | | **Novelty** | 3/10 | [LIT] Multiple 2023/24 papers already use BanglaBERT and GPT-3.5 for this. | | **Feasibility** | 8/10 | [FACT] Public datasets like SentNoB exist; fits free T4 Colab limits. | | **Depth** | 4/10 | [INFERENCE] Standard fine-tuning is now considered a baseline, not a contribution. | **Verdict: PIVOT.** A standard sentiment classifier is likely to be rejected as incremental. To salvage this, I recommend focusing on **Cross-lingual Knowledge Transfer for Low-Resource Dialects** or **LLM Uncertainty Quantification in Bangla Sarcasm**. ## Feasibility Check (Google Colab T4) | Component | Status | Detail | | :--- | :--- | :--- | | Model Size | [PASS] | 7B parameter models (4-bit quant) fit in 16GB VRAM. | | Dataset | [PASS] | SentNoB (15k samples) is public and manageable. | | Runtime | [PASS] | Fine-tuning via LoRA takes < 4 hours on T4. | **Next steps** 1. Run `/papermate "Topic funnel for Bangla NLP"` to see the 3 strongest publication-ready variants of this idea. 2. Select a specific research question to generate the **Experiment Design** and **Notebook Scaffolds**. 3. Locate the `PROJECT.md` I have initialized in your directory to track these decisions.

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

    About this skill

    Research in NLP and ML is plagued by incremental ideas that fail peer review or stall due to compute limits. PaperMate acts as a skeptical senior researcher and reviewer, identifying fatal flaws in research ideas before you invest months in implementation. It forces honest novelty testing, validates feasibility for Google Colab environments, and ensures every claim in your manuscript is grounded in a verified results ledger. ## What it does - **Idea autopsy** runs aggressive novelty checks against current literature to kill or salvage weak research questions. - **Feasibility auditing** calculates if a project is doable within specific VRAM, dataset size, and budget constraints. - **Experiment scaffolding** generates structured Google Colab notebooks with automated checkpointing and results logging. - **Results-grounded writing** drafts manuscripts using a ledger-based system to prevent hallucinated numbers or unverified citations. - **Reviewer simulation** provides harsh, multi-perspective peer reviews to catch weaknesses before submission to Elsevier journals. ## How it works 1. **Initiate reconnaissance** by providing a raw idea, a base paper, or a request for topic mining. 2. **Pass the gates** where the skill evaluates novelty (Gate 1) and compute feasibility (Gate 2) before proceeding. 3. **Execute experiments** using the generated notebook scaffolds and logging all outputs to the `ledger.jsonl`. 4. **Target journals** by analyzing aims, scope, and impact metrics of potential venues for the finalized manuscript. ## Frameworks & tools This skill is optimized for NLP and ML research using Python, Google Colab, and the Elsevier journal ecosystem. It uses local scripts for project scaffolding and manuscript linting against JSONL results ledgers. ## Why this beats prompting it yourself Generic prompts often suffer from toxic positivity, encouraging incremental ideas that reviewers will eventually reject. This skill uses a strict "kill my idea" protocol and a non-negotiable results-grounding system that makes it impossible to accidentally include unverified stats or fabricated DOIs. ## Use cases - **Thesis mining** to find a publishable NLP topic with available datasets and low compute requirements. - **Literature reconnaissance** to map the delta between your idea and the last three years of arXiv preprints. - **Protocol design** to create an ablation matrix and statistical testing plan that satisfies high-tier journal requirements. - **Manuscript hardening** to simulate a three-person reviewer panel and fix experimental gaps before submission. ## Known limitations Cannot execute code inside your Google Colab environment; you must paste results back into the session. Search capabilities are limited by the tools available to the agent.

    Requires manual data entry of experiment results into a ledger. Cannot run Colab cells directly. Novelty checks depend on the availability of web search or academic search tools.

    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

    3.0
    2 reviews
    5
    1
    4
    0
    3
    0
    2
    0
    1
    1

    This skill has been used 4 times.

    Trust & safety

    Security scan

    Security scan passed 28 days ago

    • Free with an Agensi account

    Requires

    Python

    Permissions

    Terminal / Shell
    Network Access
    Read Files
    Write Files
    Listed28 days ago

    What's inside

    Frequently Asked Questions

    PaperMate is designed to act as a cynical senior researcher, identifying fatal flaws, novelty gaps, and compute constraints in your research ideas before you waste time and resources on implementation.

    The skill generates structured Google Colab scaffolds and local scripts; you will need a Google account for experimentation and a Python environment to run the manuscript-ledger linting tools.

    PaperMate uses a 'kill my idea' protocol to aggressively compare your proposal against current arXiv and Elsevier literature to ensure your work meets the novelty threshold for high-tier journals.

    Yes, you can input your hardware specs (like VRAM limits), and the skill will audit your proposed model architecture and dataset size to confirm if the project is feasible on your setup.

    You receive the full PaperMate agent logic, a suite of local project scaffolding scripts, and the proprietary JSONL results-ledger system for grounded manuscript writing.

    PaperMate provides a multi-perspective peer review simulation that mimics the harsh scrutiny of Elsevier reviewers, identifying weak experimental gaps and citation errors before you submit.

    New in Research & Analysis