Asi claude-questions-leaderboard

Track Claude's best and worst questions in DuckDB. Use when Claude asks a notably good or bad question, or to review question quality. Triggers: leaderboard, bad question, good question, question quality.

install

source · Clone the upstream repo

git clone https://github.com/plurigrid/asi

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/plurigrid/asi "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/claude-questions-leaderboard" ~/.claude/skills/plurigrid-asi-claude-questions-leaderboard && rm -rf "$T"

manifest: skills/claude-questions-leaderboard/SKILL.md

source content

Claude Questions Leaderboard

DB:

/Users/alice/v/claude_questions_leaderboard.duckdb

Script:

python3 /Users/alice/.claude/skills/claude-questions-leaderboard/scripts/leaderboard.py

Operations

python3 scripts/leaderboard.py show [bad|good]
python3 scripts/leaderboard.py bad "question" "context" "why bad"
python3 scripts/leaderboard.py good "question" "context" "why insightful"

When to add entries

Context and why fields MUST be grounded in the actual interaction — quote what the user said, describe what Claude did and didn't do, reference the tools that were or weren't used. Do not write generic descriptions.

Bad: Claude asked a question when tools could have answered it. Claude assumed the user was wrong. Claude didn't use available skills/MCPs before asking.

Good: Claude asked something that reframed the problem, surfaced a hidden connection, or showed real understanding of intent. NOT "delivering well" — the question itself must be insightful.

Principle

Exhaust tools before asking. When you do ask, ask something worth asking.