skills.rest
wangdepin
wangdepinCOMMUNITY
Explore all skills by this author →
GitHub

experiment-management

Manages experiment lifecycle, validates HuggingFace artifact uploads, and blocks compute before red-team passes.

Updated May 12, 2026
Search Tags:#workflow-orchestration#huggingface#artifact-validation#experiment-management#red-teaming#dashboard-sync
One-click install
npx skills add https://github.com/wangdepin/raca-codex --skill experiment-management-wangdepin
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: experiment-management
Source: https://github.com/wangdepin/raca-codex/tree/main/.claude/skills/experiment-management
Command: npx skills add https://github.com/wangdepin/raca-codex --skill experiment-management-wangdepin

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill prevents wasted compute and broken experiment workflows by enforcing a consistent experiment lifecycle, state tracking, and artifact validation whenever users discuss experiments.

Core Features & Use Cases

  • Automatic experiment folder provisioning: Creates and maintains the full experiment directory structure under notes/experiments/<experiment-name>/, including flow state, logs, and readmes.
  • Lifecycle hard gates for safety and correctness: Enforces red-team completion before compute runs, and requires post-artifact validation before analysis proceeds.
  • Artifact pipeline with dashboard synchronization: Uploads artifacts to HuggingFace, verifies and validates them, updates HUGGINGFACE_REPOS.md, and triggers dashboard sync so progress is visible as it happens.
  • Works with parallel design skills: Coordinates with any concurrent design/brainstorming skill by handling infrastructure and lifecycle steps while the design skill drives the creative/conversational side.

Quick Start

Ask the agent to set up and manage an experiment for a model comparison, and it will create the experiment folder, sync the dashboard, run the required red-team preflight, and manage validated artifact uploads.

Frequently Asked Questions about experiment-management

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I enforce a red-team review gate before running compute-heavy experiment workflows?▼

To enforce red-team review before compute, experiment lifecycle management blocks compute execution until red-team preflight passes, tracking state in flow_state.json across design, redteam, canary, validate, running, review, and complete phases.

How do I structure experiment folders and track lifecycle states for research runs?▼

Experiment folder provisioning automatically creates and maintains the full directory structure under notes/experiments/<experiment-name>/, including flow state, logs, and readmes, ensuring structured experiment lifecycle tracking and validated artifact handling.

How does artifact validation and HuggingFace upload work for experiment lifecycle management?▼

Artifact validation uploads artifacts to HuggingFace, verifies and validates them, updates HUGGINGFACE_REPOS.md, and triggers dashboard synchronization so experiment progress is visible immediately during the validate and review phases.

Can I run a parallel design skill alongside experiment lifecycle and artifact validation?▼

Yes, experiment lifecycle management coordinates with concurrent design or brainstorming skills by handling infrastructure, folder provisioning, and lifecycle enforcement, while the design skill drives the creative and conversational side of the research workflow.

Why block compute execution until red-team review passes during experiment management?▼

Blocking compute until red-team review passes prevents wasted compute and broken experiment workflows by enforcing lifecycle hard gates, ensuring experiment safety and correctness before allowing expensive model comparison runs to proceed.

Related Skills

Explore More Education & Research Skills→
datamonsterrdatamonsterr

create-new-experiment

Bootstrap a reproducible autoresearch experiment folder with templates and registry updates.

Community
Intermediate
WangJie-cnWangJie-cn

experiment-manager

Track ML experiment configurations, metrics, and GPU usage.

Community
Advanced
Roeia1Roeia1
13

dashboard

Start a local server to visualize epics, stories, and execution progress.

Community
Basic
mfittkomfittko
14

dev-loop

Automate GitHub issue-to-merged PR workflows through seven deterministic phases.

Community
Advanced