When you run a command to reproduce a deep learning result, you need proof. This skill helps you capture that proof in a standardized way. It creates special output files and patch notes so others can see exactly what happened.
The skill works after you already have a plan and a command to run. It does not choose the goal or set up the environment. It just captures the evidence from your run and organizes it into reports.
You will get files like SCIENTIFIC_CHANGELOG.md and COMPARABILITY_REPORT.md. These make your work clear and easy to audit. This is great for anyone who wants to share reproducible research.
Global
mkdir -p ~/.claude/skills/minimal-run-and-auditProject
mkdir -p .claude/skills/minimal-run-and-auditSource Repository
Lark Baselarksuite/cli
Simplify your Lark Base data with tables, fields, records, and views
Lark Wikilarksuite/cli
Manage your Lark Wiki spaces, members, and documents with simple commands
Grillingmattpocock/skills
Test your plan thoroughly with clear step-by-step interview questions
Analyze Projectlllllllama/rigorpilot-skills
Safe read-only analysis of deep learning repos for understanding and risk detection
Ai Research Reproductionlllllllama/rigorpilot-skills
Faithful reproduction of deep learning research with auditable steps and minimal changes
Repo Intake And Planlllllllama/rigorpilot-skills
Scan a repo, find commands, and get a simple reproduction plan
Minimal Run And Auditlllllllama/rigorpilot-skills
Capture evidence from deep learning runs and create standardized audit reports
Obsidian Vaultmattpocock/skills
Search create and organize notes with wikilinks and index notes easily