Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/mxslr/mlcraft/ml-research-methodologynpx skills add mxslr/mlcraft --skill ml-research-methodologygit clone --depth 1 https://github.com/mxslr/mlcraftWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/mxslr/mlcraft/ml-research-methodology)<a href="https://agentmods.dev/skills/mxslr/mlcraft/ml-research-methodology"><img src="https://agentmods.dev/badge/skills/mxslr/mlcraft/ml-research-methodology.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00127 | $0.01450 |
| Opus 5 | $0.00063 | $0.00725 |
| Sonnet 5 | $0.00025 | $0.00290 |
| Haiku 4.5 | $0.00013 | $0.00145 |
Grade A, and why
ml-research-methodology scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 71 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ML Research-Engineer Methodology (Orchestrator)
Act as a senior AI/ML engineer and researcher, not a code monkey. The goal is the correct, defensible result - not the highest-looking number. Work through these phases in order. Announce the phase you are in.
Phase 0 - Framing (before anything)
- What is the problem type? (classification / detection / segmentation / regression / forecasting / generation / retrieval)
- What metric actually matters for the use case? (e.g., recall/sensitivity for cancer screening; mAP for detection; calibration for decision support). Do not default to accuracy.
- What is the deployment context? (offline benchmark vs a doctor/user uploading one input, which affects single vs multi-input, latency, explainability).
- What deliverable target does the user want: a Kaggle notebook, a Google Colab notebook, or a local GPU run? Invoke
notebook-deliveryto decide this and to follow the required notebook style (short simple explanations, professional research tone, no em-dash, no emoji). Ask if unstated. - State the honest difficulty: is this an easy or a genuinely hard problem?
Phase 1 - Literature review (research FIRST)
Invoke literature-review (and the paper-researcher agent). Establish:
- SOTA methods for this exact task + dataset, and the realistic, leakage-free ceiling.
- Which reported numbers are inflated (ROI-crop instead of whole input, leaky splits, tiny test sets). Never promise to beat inflated claims.
Phase 2 - Understand the data, then audit leakage
First invoke dataset-profiling to inspect the ACTUAL data (read local files, or emit a Kaggle profiling cell to run and paste back). Then invoke data-rigor-and-leakage BEFORE training. Wrong splits invalidate everything.
- Split by group/patient/temporal as appropriate; prefer the dataset's official split.
- Verify labels, class balance, and that preprocessing is identical train/val/test.
Phase 3 - Method selection: route to a domain skill
Pick the approach grounded in Phase 1 + the actual dataset, not hype. Route by task signal:
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 71 lines · 127 tokens per session scan A 6b61cd910638
ml-research-methodology is a skill published in the GitHub repository mxslr/mlcraft (8 stars, last pushed 1mo ago), licensed MIT. It adds 127 tokens to every session and 1,450 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
reinforcement-learning
Reinforcement Learning best practices for Python using modern libraries (Stable-Baselines3, RLlib, Gymnasium). Use when: Implementing RL algorithms (PPO, SAC, DQN, TD3, A2C) Creating custom Gymnasium environments Training, debugging, or evaluating RL agents Setting up hyperparameter tuning for RL Deploying RL models…
ml-for-aec
Computer vision for buildings, image-to-floorplan, generative ML models, performance prediction, structural analysis ML, energy prediction, natural language to design, and point cloud ML for AEC computational design.
ml-for-aec
Computer vision for buildings, image-to-floorplan, generative ML models, performance prediction, structural analysis ML, energy prediction, natural language to design, and point cloud ML for AEC computational design.
physicsnemo-discover
Official NVIDIA-authored guidance for navigating PhysicsNeMo — pick the model, datapipe, or example for a SciML/AI4Science task (surrogates, forecasting, downscaling, physics-informed, inverse, generative). Points at existing files via live repo search; never writes code. Do NOT use for installation or environment…
torch-geometric
Graph Neural Networks (PyG). Node/graph classification, link prediction, GCN, GAT, GraphSAGE, heterogeneous graphs, molecular property prediction, for geometric deep learning.
esm
Comprehensive toolkit for protein language models including ESM3 (generative multimodal protein design across sequence, structure, and function) and ESM C (efficient protein embeddings and representations). Use this skill when working with protein sequences, structures, or function prediction; designing novel…