Prompt improver

Skill SamyakJhaveri/loam/cultivation/marketplace/helpers/skills/prompt-improver

Copier template that bootstraps AI-agent-optimized project setups: layered context routing, curated skills, and an enforced validation gate — synced across projects.

Install

npx -y skills add SamyakJhaveri/loam --skill prompt-improver

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when invoked automatically by the UserPromptSubmit hook after a prompt is flagged as vague. Researches context, then generates 1-6 targeted clarifying questions before executing. Do NOT invoke this manually — it is triggered by the hook in .claude/hooks/improve-prompt.py.

SKILL.md

2.2 KB, 419 tokens by cl100k_base, as published. Nobody here has run it

Prompt Improver Skill

You are invoked because the UserPromptSubmit hook has determined the user's prompt lacks sufficient clarity to execute well. Your job is to research context first, then ask smart questions — not the other way around.

Phase 1 — Research (NEVER skip this)

Before generating any questions, gather real context:

Check the full conversation history — has this topic come up before? Is there already enough context?
If the request involves code: read relevant files, scan the directory structure, understand the stack
If the request involves a task: understand what already exists, what's already been tried
If you can fully infer the user's intent from context alone — skip to Phase 4 immediately

Phase 2 — Generate Questions (only if genuinely needed)

Create 1–6 questions. Rules:

Every question must be grounded in what you actually found in Phase 1
Never ask about something you can already infer
Each question should be multiple-choice (3–4 options max) — no open-ended text boxes
Questions must reduce ambiguity, not pad the conversation
If 1 question resolves the ambiguity, only ask 1

Phase 3 — Clarification

Use the AskUserQuestion tool to present your questions. Each option should:

Be concrete and specific (not vague like "Option A / Option B")
Describe the trade-off or implication where relevant
Include an "Other / I'll describe" option

Phase 4 — Execute

Proceed with the original request using:

The user's original intent
What you learned in Phase 1
The user's answers from Phase 3 (if any questions were asked)

Do not re-ask anything the user has already answered. Do not summarise what you're about to do — just do it.

Bypass

If the user prefixes their message with *, the hook skips evaluation entirely and passes the prompt straight through. Useful for urgent or already-detailed prompts.