tpu_performance_autoresearch_wiki: Skill for Claude Code

.claude/skills/create-experiment/SKILL.md

create-experiment is a skill for Claude Code from vlasenkoalexey/tpu_performance_autoresearch_wiki. It costs 90 tokens per session (2,972 once invoked), scanned A, original, MIT.

A setup procedure for adding a new model family to an autoresearch project. A model family is a related group of model versions; the procedure creates its folders and planning files.

In plain words
What is it for?
Use it to create model documentation under the experiments directory and optionally add planning stubs for different execution lanes.
Why use it?
It avoids starting experiments with missing structure or unclear choices about model sizes, execution methods, hardware, and sequence length.

Skill for Claude Code

Written for Claude Code: installed under .claude/. Also seen: names the AskUserQuestion tool.

This is vlasenkoalexey/tpu_performance_autoresearch_wiki's own configuration. It tells Claude Code how to work on tpu_performance_autoresearch_wiki itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything tpu_performance_autoresearch_wiki configures →

Needs its repository: it reads a path above its own folder, which exists only inside the repository. The line is Model-family specialization for the **<DISPLAY_NAME>** family. Inherits from `../program.md`; overrides the sections below..

Reuse

Borrowing it

Nothing to install: this file belongs to vlasenkoalexey/tpu_performance_autoresearch_wiki. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/vlasenkoalexey/tpu_performance_autoresearch_wiki/main/.claude/skills/create-experiment/SKILL.md
Clone the repo
git clone --depth 1 https://github.com/vlasenkoalexey/tpu_performance_autoresearch_wiki

Made for: Claude Code.

Wrote 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.

agentmods badge for create-experiment

README.md
[![agentmods](https://agentmods.dev/badge/skills/vlasenkoalexey/tpu_performance_autoresearch_wiki/create-experiment/github.svg)](https://agentmods.dev/skills/vlasenkoalexey/tpu_performance_autoresearch_wiki/create-experiment)
Your own site
<a href="https://agentmods.dev/skills/vlasenkoalexey/tpu_performance_autoresearch_wiki/create-experiment"><img src="https://agentmods.dev/badge/skills/vlasenkoalexey/tpu_performance_autoresearch_wiki/create-experiment/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for create-experiment

Your own site · 80×15
<a href="https://agentmods.dev/skills/vlasenkoalexey/tpu_performance_autoresearch_wiki/create-experiment"><img src="https://agentmods.dev/badge/skills/vlasenkoalexey/tpu_performance_autoresearch_wiki/create-experiment.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 90 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,972 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector warn 7 Sept 2026
SkillSpector: 2 findings, up to medium

These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →

  • medium Excessive Agency · line 38
    Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
    Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
  • medium Excessive Agency · line 111
    Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
    Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
How audits are shown
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00090 $0.02972
Opus 5 $0.00045 $0.01486
Sonnet 5 $0.00018 $0.00594
Haiku 4.5 $0.00009 $0.00297

Measured 11d ago against content hash 482127a740e8, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade A, and why

create-experiment 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 11d 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.

.claude/skills/create-experiment/SKILL.md · 308 lines

How it starts

The opening of the file, as written. The whole thing — 308 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are bootstrapping a new model family for the autoresearch optimization loop. Follow this sequence — ask questions, then create files. Do NOT silently write files without confirming with the user.

Step 1 — Gather model details via AskUserQuestion

Ask these questions in order (some may be combined into a single AskUserQuestion call with multiple questions):

Question 1: "Model folder slug?" (free-text input expected)

  • This becomes the folder name under wiki/experiments/. Convention: lowercase, underscores, no spaces.
  • Examples: gemma4, gemma5, qwen3_8B, llama4_70B.
  • The full folder will be wiki/experiments/<slug>_autoresearch_optimization/.

Question 2: "Model display name?" (free-text)

  • Human-readable name for documentation. Used in the model-level program.md title.
  • Examples: "Gemma 4", "Gemma 5", "Qwen3 8B", "Llama 4 70B".

Question 3: "Which lanes?" (multi-select)

  • Options: tpu (PyTorch eager on TPU), jax (JAX + Flax NNX), torchax (torchax bridge), maxtext (MaxText). User can also type a custom lane name.
  • The lane folders will be created under wiki/experiments/<slug>_autoresearch_optimization/.

Question 4: "Model sizes?" (free-text, comma-separated)

  • Examples: "3B, 24B", "1B, 8B, 70B", "125M, 1.3B".
  • These populate the sizes: field in model page frontmatter and the architecture description.

Question 5: "Target hardware variants?" (free-text, comma-separated)

  • Examples: "v5p-8, v5p-32, v6e-16", "v6e-8, v6e-32".
  • These populate the hardwares: field. Used to build the variant matrix on the model page.

Question 6: "Target sequence length for MFU optimization?" (number input)

  • Default: 8192.
  • Examples: 8192 (most modern LLMs), 4096, 16384.

Question 7 (optional): "Architecture invariants?" (free-text, multiline)

  • Things that must NOT change without approval. Examples: "8 parallel tracks, 4 experts top-2 routing", "GLU activation, fused attention", etc.
  • Skip if user has nothing specific yet — they can fill in later.

Question 8 (optional): "Trainer entry-point pattern?" (free-text)

  • Default: torchtitan.experiments.<lane>.<slug>.train_minimal (matches the existing convention).
  • User can override if their model uses a different entry pattern.

Read the full file on GitHub · 308 lines

Changes

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.

  1. 11d ago First seen · 308 lines · 90 tokens per session scan A 482127a740e8

Subscribe to this mod's changes

create-experiment is a skill published in the GitHub repository vlasenkoalexey/tpu_performance_autoresearch_wiki (55 stars, last pushed 7d ago), licensed MIT. It adds 90 tokens to every session and 2,972 once invoked, about $0.0005 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 2026-08-30.

Related

Other skills, from other repositories

ml_inference_optimization

This document describes the Leeroopedia MCP tools available during the with-KB benchmark run. It is a standalone reference and is not fed to the agents automatically.

Leeroo-AI/leeroopedia-mcp · 0 tokens

leeroopedia-mcp

Use Leeroopedia MCP to fetch grounded ML/AI best practices, build and review ML plans, debug failures, verify code/math correctness, and expand KB citations via getpage.

Leeroo-AI/leeroopedia-mcp · 42 tokens

ml-research-lab

Machine-learning research loop for dataset curation, fine-tuning, evaluation, inference deployment, experiment tracking, and model explainability. Use when working on ML experiments, training data, model benchmarks, RunPod/GPU runs, classifier quality, vLLM/GGUF serving, SHAP-style model explanations, or…

AnastasiyaW/codex-claude-code-config · 96 tokens

experiment-audit

Use this skill for scientific and ML-research reasoning work — evaluating experimental claims, auditing training runs or ablations, checking whether a statistical claim holds up, assessing reproducibility, reconciling contradictory results, reviewing a paper's methodology or results section, writing reviewer-style…

SreeDharshan-GJ/experiment-audit · 276 tokens

midmem-research-tracker

Keep the repo's RESEARCH.md ledger current from the research ingestions already in the MidMem store: for every unevaluated weekly research ingestion, evaluate each paper's feasibility and applicable improvement for MidMem, and record it as Paper (cite + link) · Finding (the empirical claim that matters) · Decision…

U4CSolutions/MidMem · 160 tokens

sematyp

SemaTyP combines two data sources into a knowledge graph for drug discovery / repositioning.

QSong-github/DrugClaw · 0 tokens