ask-less-execution

name	ask-less-execution
description	Make the agent continue work without asking unnecessary questions. Enforce the ask-less policy from CLAUDE.md. Trigger on /ask-less-execution, or when the agent is asking too many clarifying questions instead of discovering answers.
user-invocable	true
allowed-tools	Read, Grep, Glob, Bash

Ask-less Execution

Stop asking questions whose answers are discoverable. Use this skill when the agent is interrupting flow with unnecessary clarifying questions.

When to Ask

Ask only when:

• Behavior has multiple valid interpretations and none is clearly conventional.
• A destructive or risky operation is needed (delete, force-push, format, mass edit).
• Security, authentication, payment, or database migration is involved.
• UI or branding decisions cannot be inferred from existing code.
• External accounts, secrets, or permissions are required.

When NOT to Ask

Do not ask when:

• The answer is discoverable in the repo (package manager, config files, test commands, API routes, existing patterns).
• A safe, reversible default assumption exists.
• Similar patterns can be found in existing code.
• A reasonable convention is well-established in the ecosystem.

Procedure

1. Before asking, search the repo. Read package.json, config files, existing code.
2. If the answer is still ambiguous and the impact is low, make a reasonable assumption and note it.
3. If the decision is genuinely ambiguous AND impactful, ask exactly one concise question.
4. After receiving an answer, proceed without further confirmation on routine steps.

Operating Rules

• Search first, ask second.
• Default to conventional assumptions (common package managers, standard test frameworks, conventional project layouts).
• Note assumptions in the response so the user can correct them.
• If a user says "stop asking" or "just do it", escalate to this skill.

Experience Log

This skill learns from experience. Mechanism:

• Read .claude/experience/ask-less-execution.md at skill start to benefit from past lessons.

• Write to .claude/experience/ask-less-execution.md after use if you learned something new.

• Format: YYYY-MM-DD: <lesson> — one line per entry.

• Evolve: If the same issue repeats 3+ times, update this SKILL.md itself.

This skill improves over time. After each use, append lessons learned.

Record format: YYYY-MM-DD: <lesson> Read this at skill start to apply past lessons.