stash-name-gen_skill

Overview

This skill generates meaningful git stash names from the actual changes in your working tree so you stop using vague labels like "WIP". It inspects staged and unstaged diffs, queries a language model to produce a concise description, and then runs git stash with that message. One command, no configuration needed.

How this skill works

The tool runs git diff (and staged diffs) to collect a summary of changed files and hunks, sends that summary to a small GPT model to create a human-readable stash name, and then executes git stash push -m with the generated message. It also supports a dry-run mode to preview the name without creating the stash.

When to use it

• Before switching branches to capture current work with a clear label
• Prior to pulling or merging to save in-progress changes
• When experimenting or prototyping so your stashes are searchable later
• Any time you would normally run git stash and want a descriptive message
• When cleaning your working tree but planning to resume work soon

Best practices

• Run the command frequently—short, descriptive stashes are easy to revisit
• Use --dry-run to preview and confirm long or ambiguous names
• Keep stashes short and actionable; include the intent or affected component
• Avoid hoarding stashes—apply or drop them once work is resumed or merged
• Ensure your OPENAI_API_KEY is set if using the hosted model integration

Example use cases

• Stash changes that fix a bug in the auth flow before switching to a hotfix branch
• Save experimental UI tweaks with a name like "refactor: sidebar state + fix layout"
• Capture partial feature work before pulling upstream changes
• Store quick prototype work so you can iterate without losing context
• Create readable stash messages across a team to speed retrieval