dont-ask-me

Goal

Enable a "no-ask" mode from invocation until the end of the session — never use /ask-me, ask_user_question, or any selection-style prompts, even when global_rules.md or other skills instruct /ask-me. Replace them with /follow-your-suggestion and work incrementally and safely without skipping steps, until the job is fully complete via /loop-until-complete

Scope

• Use when the user says dont-ask-me, dont ask me, do not ask, no need to ask, proceed, or no ask, or when a workflow such as /ship enables this mode in dont-ask-me mode
• Works as a session mode: once enabled it applies to every instruction in the same session, including global_rules.md and any skills that reference /ask-me
• Not a store of requirements or preferences — if a prompt contains a question that must be answered before starting, answer it with a safe default via /follow-your-suggestion instead of asking

Execute

1. Activate Session Mode

Goal: turn on no-ask mode for the rest of the session

1. Record that this session is in dont-ask-me mode from now until the session ends
2. Set a temporary rule: everywhere global_rules.md, a skill, or a workflow says to run /ask-me or ask_user_question → replace it immediately with /follow-your-suggestion
3. If a prompt has an outstanding question → answer it with a safe default and record the decision; do not ask back

2. Evaluate Context

Goal: confirm it is safe to proceed

1. Read context, memory, project conventions, and AGENTS.md
2. Check that the choice has a clear and safe outcome
3. If still unclear → pick a safe default via /follow-your-suggestion instead of /ask-me or /ask-project-requirement

3. Pick Safe Default

Goal: choose the simplest and safest option

1. Pick the option with the lowest risk that is reversible
2. Ground the choice in best practices and project conventions
3. If several options are equal, take the simplest and most reversible one

4. Proceed Step By Step

Goal: work incrementally and safely without skipping steps

1. Proceed one step at a time in order; never skip a step the workflow defines
2. Change one small point at a time and check the result before continuing — never bundle several risky changes in one round
3. Do not use ask_user_question, do not show selection prompts, do not call /ask-me or /ask-again
4. Use /follow-your-suggestion to apply the suggestion instead of asking for confirmation
5. Record every decision in the report so it stays traceable

5. Loop Until Complete

Goal: repeat until the work is fully complete

1. Run /loop-until-complete to re-check until the criteria pass or a good-enough point is reached
2. If a problem appears along the way → fix it point by point with /resolve-errors and keep looping; do not stop mid-way to ask
3. Stop only when fully complete, when hitting max iterations, or when a genuinely dangerous action is found

6. Report

Goal: report results and next steps

1. Summarize what was done, why nothing was asked, and which defaults were chosen
2. State that dont-ask-me mode is still active for the next tasks in the session
3. Run /suggest-next-action to recommend the next step without asking

Rules

1. Session Override

• While dont-ask-me is active → override every /ask-me and ask_user_question in global_rules.md and all skills until the session ends
• /ask-me, /ask-again, /ask-project-requirement, and pick-bestest calls made by other workflows → run /follow-your-suggestion with a safe default instead
• The user can always cancel the mode with an explicit command such as "start asking again" or "cancel dont-ask-me"

2. No Prompt Discipline

• Never use ask_user_question while in dont-ask-me mode
• Never show prompts offering choices
• Never call /follow-your-suggestion to ask for confirmation — use it only to apply suggestions
• /continue and /loop-until-complete may be used immediately as normal

3. Safe Stepwise Execution

• Work incrementally, one step at a time in workflow order — never skip steps
• Fix incrementally, point by point, and verify before continuing
• Never take dangerous or irreversible actions — pick the safest path and report instead of asking

4. Uncertainty Handling

• If doubt arises mid-work → stop that step, report the status, then choose a safe default to continue with other parts
• If no safe option exists at all → stop and report without asking

Expected Outcome

• No /ask-me or user-facing prompts at all until the session ends
• Every point that once required a question is decided via /follow-your-suggestion with a traceable safe default
• Work proceeds safely step by step, never skipping steps, and loops until complete via /loop-until-complete
• Fewer unnecessary questions without added risk, and the user never has to answer repeated questions