lararosekelley/dotfiles/discuss
Claude Code and agent skill by lararosekelley/dotfiles
- community
Checking that the SKILL.md file is still in the repository.
Shape an idea into an actionable plan through short rounds of informed questions. Use when the user invokes /discuss or explicitly asks for an interactive planning conversation before implementation. Do not trigger for a simple question or a request to start building an already agreed change.
Install
npx vskill@latest install lararosekelley/dotfiles/discussSKILL.md
Discuss
Help the user settle the decisions needed to implement, without implementing. Follow current repository instructions and keep investigation read-only.
Help dispatch
If the entire argument is help, --help, or -h, print Usage and stop before investigating. Otherwise use the invocation or forwarded arguments as the topic.
Usage
/discuss [idea, problem, or plan|help]
Inspect relevant context, then ask up to three focused questions per round, each with a recommended/default answer and a short reason. With no topic, use a clear topic from this conversation or ask what you want to discuss.
Reply naturally, redirect the discussion, or say stop to return to normal conversation. When the decisions are sufficient, get a concise plan. Approval of the plan does not start implementation; explicitly ask to build when ready.
Investigate before asking
- Read relevant code, instructions, and docs using the project's search guidance.
Ask only questions that cannot be answered by inspecting the project. Keep research bounded to the next decision rather than surveying everything first.
- Identify the goal, observable success criteria, constraints, and what must not
change. Use conversation context instead of repeating answered questions.
- Surface the next unresolved decision, assumption, dependency, or tradeoff.
Resolve prerequisites before dependent choices. Consider doing nothing or a simpler solution when it meets the goal; do not assume a new feature is needed.
Ask and converge
Ask no more than three questions per turn. For each, give a concrete recommendation and a brief reason. Make alternatives materially different; do not invent choices to fill a menu. Use the harness's question UI when available and useful, or plain text otherwise. Wait for answers before the next round.
Keep agreed decisions in conversation, not new tracking files. Treat uncertainty as uncertainty and challenge a premise when evidence warrants it. Do not keep asking optional polish questions once scope, behavior, and verification are clear enough to proceed. If choices require an experiment, propose it and ask before performing work beyond read-only investigation.
Finish or stop
Summarize agreed decisions, meaningful open questions, the recommended approach, and the next step. Keep the plan in chat unless the user asks for a file.
Do not edit code, install dependencies, create branches, file issues, or publish anything as part of discussion. "Yes" to a proposed design settles that decision; it does not authorize implementation or Git operations. If the user explicitly asks to implement, leave discussion mode and follow the normal approval rules.
On stop, end the questioning loop. Give a short summary only if it is useful, then return to normal conversation without beginning implementation.
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingcontent/.claude/skills/discuss/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/lararosekelley/dotfiles/discuss)