# Go > Accept the recommendation from the most recent A/B/C option set and execute it without re-asking — for "按你推荐的来" / "go with your pick", when the user glanced at the options and wants the recommended branch executed immediately. - Skill: `shenxingy/go-3` (Agent Skill) - Install (CLI): `npx skillmds@latest add shenxingy/go-3` - Raw SKILL.md: https://api.skillmd.com/api/skills/shenxingy/go-3/raw - Safety review: pending (external: skill-scanner PASS, skillspector WARNING) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: shenxingy (https://skillmd.com/u/shenxingy) - Updated: 2026-09-21 - Page: https://skillmd.com/skills/shenxingy/go-3 --- # Clade for Codex This workflow runs **directly in Codex**. Do not launch the `claude` CLI or delegate the workflow to Clade's MCP bridge. Codex compatibility rules: - Plugin skills are namespaced. Invoke this workflow explicitly as `$clade:go`; a bare `$name` does not select the installed Clade plugin. - Read the nearest `AGENTS.md` files for repository instructions. If a project has only `CLAUDE.md`, treat it as legacy project guidance and read it too. - Store new Clade working state under `.clade/` (or `~/.clade/` for personal state). Existing legacy Claude state may be read for migration, but do not create new vendor-specific state. - A `/skill-name` reference means the corresponding Codex `$clade:skill-name` plugin skill, or the same workflow invoked naturally when explicit skill invocation is not available. - Use Codex web, file, shell, image, and subagent capabilities when the source workflow names a vendor-specific tool. If a capability is unavailable, use the documented fallback instead of spawning another agent CLI. - Paths such as `/...` are relative to the installed Clade plugin containing this `SKILL.md`; resolve that root before invoking a helper. ## Canonical Clade workflow name: go trigger: user says "按推荐的来" / "go with your pick" after you offered enumerated options completion-status: DONE | DONE_WITH_CONCERNS | BLOCKED | NEEDS_CONTEXT The user just saw a set of options you offered (numbered issues with A/B/C sub-options, or a recommendation block with alternatives). They don't want to re-read them — they want the one YOU recommended to be executed now. ## Execution ### Step 1: Locate the most recent options block Scan the prior assistant messages in this conversation for the most recent message that: - Enumerated 2+ options (A/B/C, 1/2/3, "Option 1/Option 2"), OR - Presented a "Top pick" / "Recommended" / "Default" alongside alternatives If multiple option sets exist in recent history, use the **most recent one that is still actionable** — ignore ones already resolved by subsequent messages. ### Step 2: Identify the recommendation Look for an explicit recommendation marker: - Words like "recommend", "推荐", "I'd go with", "Top pick", "Default:", "Suggest" - A pick you marked with `★`, `▶`, bold, or placed first as "the one to pick" ### Step 3: Confirm and execute **If the recommendation is unambiguous:** - State in one line: `→ going with {option label}: {short description}` - Execute immediately — no re-asking, no re-explaining tradeoffs - Proceed to the full task the options were gating **If options existed but no clear recommendation was flagged:** - Surface the top 2 in a single line each with tradeoffs - Ask: "No clear default in my last message — {A} or {B}?" - Do NOT execute until the user picks **If no recent options block exists (within the last ~5 assistant turns):** - Respond: "Nothing pending to pick — what would you like me to go with?" - Stop. ## Rules - **One sentence of confirmation, then act.** This skill exists because the user wants speed. - **Don't re-derive** the recommendation — trust what you already decided. If you said "I recommend A" 30 seconds ago, pick A now. - **Don't expand scope.** `/go` executes the specific option; it doesn't also kick off related work you didn't offer. - **If the recommended option has side effects that need confirmation** (file deletion, `git reset --hard`, force-push, migrations, `.env` edits): confirm once before executing. `/go` does not override the destructive-action rule in AGENTS.md. - **Multiple numbered issues with their own A/B/C** (e.g. from Plan Mode): pick the recommended letter for each numbered issue and execute them in order. Mention the combined choice in one line (e.g. `→ going with 1A, 2B, 3A`). ## Completion Status - ✅ **DONE**: Identified recommendation, executed it, produced the expected output. - ⚠ **DONE_WITH_CONCERNS**: Executed but noticed the recommendation now looks weaker than when first given — flag that. - ❌ **BLOCKED**: No clear recommendation or no recent options — asked user for clarification. - ❓ **NEEDS_CONTEXT**: Recommendation requires a destructive action that needs explicit approval. ## Additional skill reference # Go Skill The user just saw a set of options you offered (numbered issues with A/B/C sub-options, or a recommendation block with alternatives). They don't want to re-read them — they want the one YOU recommended to be executed now. ## Execution ### Step 1: Locate the most recent options block Scan the prior assistant messages in this conversation for the most recent message that: - Enumerated 2+ options (A/B/C, 1/2/3, "Option 1/Option 2"), OR - Presented a "Top pick" / "Recommended" / "Default" alongside alternatives If multiple option sets exist in recent history, use the **most recent one that is still actionable** — ignore ones already resolved by subsequent messages. ### Step 2: Identify the recommendation Look for an explicit recommendation marker: - Words like "recommend", "推荐", "I'd go with", "Top pick", "Default:", "Suggest" - A pick you marked with `★`, `▶`, bold, or placed first as "the one to pick" ### Step 3: Confirm and execute **If the recommendation is unambiguous:** - State in one line: `→ going with {option label}: {short description}` - Execute immediately — no re-asking, no re-explaining tradeoffs - Proceed to the full task the options were gating **If options existed but no clear recommendation was flagged:** - Surface the top 2 in a single line each with tradeoffs - Ask: "No clear default in my last message — {A} or {B}?" - Do NOT execute until the user picks **If no recent options block exists (within the last ~5 assistant turns):** - Respond: "Nothing pending to pick — what would you like me to go with?" - Stop. ## Rules - **One sentence of confirmation, then act.** This skill exists because the user wants speed. - **Don't re-derive** the recommendation — trust what you already decided. If you said "I recommend A" 30 seconds ago, pick A now. - **Don't expand scope.** `/go` executes the specific option; it doesn't also kick off related work you didn't offer. - **If the recommended option has side effects that need confirmation** (file deletion, `git reset --hard`, force-push, migrations, `.env` edits): confirm once before executing. `/go` does not override the destructive-action rule in AGENTS.md. - **Multiple numbered issues with their own A/B/C** (e.g. from Plan Mode): pick the recommended letter for each numbered issue and execute them in order. Mention the combined choice in one line (e.g. `→ going with 1A, 2B, 3A`). ## Completion Status - ✅ **DONE**: Identified recommendation, executed it, produced the expected output. - ⚠ **DONE_WITH_CONCERNS**: Executed but noticed the recommendation now looks weaker than when first given — flag that. - ❌ **BLOCKED**: No clear recommendation or no recent options — asked user for clarification. - ❓ **NEEDS_CONTEXT**: Recommendation requires a destructive action that needs explicit approval. ## Delivery completion If this workflow changes files or external state: - Inspect the real final state before responding, including `git status` for a repository task. - Never report `DONE` while task-owned changes are uncommitted. Use or continue `$clade:delivery` and create a repository-compliant checkpoint or preserve the work when committing is unavailable. - When the user request or trusted repository policy makes publication, deployment, or live verification part of the task, do not silently downgrade the result to local-only work. - If a required delivery transition lacks authority, credentials, a destination, or reachable external state, report `BLOCKED` or `NEEDS_CONTEXT` rather than appending a "not committed/pushed/deployed" caveat after `DONE`.