Token导航 LogoToken导航TokenDH.com
研究检索操作浏览器github未标认证来源可访问许可证需确认审计未展示

pm%3aapprovepm%3a 批准

Agent Skill

pm%3aapprove 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

250

周安装

10

GitHub Stars

公开资料未说明

下载量

81
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:pm%3aapprove(pm%3a 批准)
来源仓库:https://github.com/cloudvoyant/codevoyant
仓库路径:skills/pm%3Aapprove
安装命令:
npx skills add https://github.com/cloudvoyant/codevoyant --skill pm:approve
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/cloudvoyant/codevoyant --skill pm:approve

简介

用于查找、检索和筛选相关信息以支持审批决策。

  • 适合根据关键词快速定位待审内容或背景资料。
  • 使用时需明确审批范围和筛选条件。适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。
  • 安装前建议确认权限范围和维护状态。
  • pm%3aapprove 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Compatibility: AskUserQuestion falls back to numbered list on non-Claude-Code platforms.

Skill Requirements

command -v npx >/dev/null 2>&1 || echo "MISSING: npx"

Critical Rules

  • Always run pm:review before promoting — do not skip
  • The draft in .codevoyant/roadmaps/ remains after promotion (source of truth for history)
  • Research artifacts from .codevoyant/explore/{slug}/ and .codevoyant/plans/{slug}/research/ are both copied flat into docs/product/roadmaps/{slug}/research/
  • Linear sync is always optional and always last
  • Never force-overwrite an existing committed roadmap without user confirmation
  • pm:approve may create or update Linear initiatives only. Never create Linear projects — that is em:approve's responsibility.

Step 0: Parse arguments

SLUG="${1:-}"
LINEAR_SYNC=false; LINEAR_URL=""; SILENT=false
if [[ "$*" =~ --push ]]; then
  LINEAR_SYNC=true
  # Capture optional URL immediately following --push
  if [[ "$*" =~ --push[[:space:]]+(https://linear\.app/[^[:space:]]+) ]]; then
    LINEAR_URL="${BASH_REMATCH[1]}"
  fi
fi
[[ "$*" =~ --silent ]] && SILENT=true

Detect artifact type

# Detect artifact type from SLUG or interactive selection
# Roadmap drafts: .codevoyant/roadmaps/*.md
# PRD drafts:     .codevoyant/prds/{slug}/{slug}.md
if [ -n "$SLUG" ] && ls .codevoyant/prds/"$SLUG"/"$SLUG".md 2>/dev/null; then
  ARTIFACT_TYPE="prd"
elif [ -n "$SLUG" ] && ls .codevoyant/roadmaps/*"$SLUG"*.md 2>/dev/null; then
  ARTIFACT_TYPE="roadmap"
else
  ARTIFACT_TYPE=""  # ask in Step 1
fi

If ARTIFACT_TYPE is empty, ask:

AskUserQuestion:
  question: "What do you want to approve?"
  header: "Artifact type"
  options:
    - label: "Roadmap — .codevoyant/roadmaps/"
    - label: "PRD — .codevoyant/prds/"

Set ARTIFACT_TYPE to "roadmap" or "prd" based on the user's choice.

If ARTIFACT_TYPE is "roadmap": proceed with the Roadmap flow (Steps 1–6 below).

If ARTIFACT_TYPE is "prd": skip to the PRD flow (Steps 1P–6P below).


Roadmap Flow

Step 1: Locate draft

If SLUG provided, resolve to .codevoyant/roadmaps/{SLUG}.md or the most recent file matching *{SLUG}*.

If no SLUG, list files in .codevoyant/roadmaps/ sorted by modification time and ask:

AskUserQuestion:
  question: "Which draft roadmap do you want to approve?"
  header: "Draft"
  options:
    - label: "Most recent draft"
    - label: "I'll specify the filename"

Read the selected roadmap. Set DRAFT_PATH and FILENAME.

Parse DATE and TIMESCALE from FILENAME:

# e.g. "260322-half-roadmap.md" → DATE=260322, TIMESCALE=half
if [[ "$FILENAME" =~ ^([0-9]{6})-([a-z]+)-roadmap ]]; then
  DATE="${BASH_REMATCH[1]}"
  TIMESCALE="${BASH_REMATCH[2]}"
else
  DATE=$(date +%y%m%d)
  TIMESCALE="roadmap"
fi
COMMIT_DIR="docs/product/${DATE}-${TIMESCALE}"

Step 2: Run pm:review

Run /pm:review on the draft. If critical issues are found, surface them and ask:

AskUserQuestion:
  question: "pm:review found critical issues. How do you want to proceed?"
  header: "Review result"
  options:
    - label: "Fix issues first (use pm:update)"
    - label: "Approve anyway — I'll address issues later"
    - label: "Cancel"

If fix or cancel, stop here.

Step 3: Confirm promotion

COMMIT_DIR was set in Step 0 (e.g. docs/product/260322-half/).

Check if COMMIT_DIR already exists. If it does, warn:

AskUserQuestion:
  question: "A committed roadmap already exists at {COMMIT_DIR}. Overwrite?"
  header: "Overwrite?"
  options:
    - label: "Yes — overwrite"
    - label: "Save as new version (add -v2 suffix to dir name)"
    - label: "Cancel"

Ask for final confirmation:

AskUserQuestion:
  question: "Promote draft to {COMMIT_DIR}?"
  header: "Confirm promotion"
  options:
    - label: "Promote"
    - label: "Cancel"

Step 4: Promote

mkdir -p "{COMMIT_DIR}"
cp "$DRAFT_PATH" "{COMMIT_DIR}/roadmap.md"

# Sources: .codevoyant/explore/{SLUG}/ and .codevoyant/plans/{SLUG}/research/ (if present)
EXPLORE_DIR=".codevoyant/explore/{SLUG}"
RESEARCH_DIR=".codevoyant/plans/{SLUG}/research"
if { [ -d "$EXPLORE_DIR" ] && [ "$(ls -A $EXPLORE_DIR 2>/dev/null)" ]; } || \
   { [ -d "$RESEARCH_DIR" ] && [ "$(ls -A $RESEARCH_DIR 2>/dev/null)" ]; }; then
  mkdir -p "{COMMIT_DIR}/research"
  [ -d "$EXPLORE_DIR"  ] && cp "$EXPLORE_DIR/"*.md  "{COMMIT_DIR}/research/" 2>/dev/null
  [ -d "$RESEARCH_DIR" ] && cp "$RESEARCH_DIR/"*.md "{COMMIT_DIR}/research/" 2>/dev/null
fi

Update agent-kit status:

npx @codevoyant/agent-kit plans update-status --name "{DATE}-{TIMESCALE}" --status Approved

Report: "Roadmap promoted to {COMMIT_DIR}/roadmap.md."

Step 5: Linear sync (optional)

If --push flag not passed, ask:

AskUserQuestion:
  question: "Sync this roadmap to Linear?"
  header: "Linear sync"
  options:
    - label: "Yes — create a new Linear initiative"
    - label: "Yes — use an existing initiative (I'll provide the URL)"
    - label: "No — skip Linear sync"

If "use an existing initiative", ask:

AskUserQuestion:
  question: "Paste the Linear initiative URL:"
  header: "Initiative URL"
  freeform: true

Set LINEAR_URL to the provided value.

If not syncing, skip to Step 6.

5a: Resolve the initiative

If LINEAR_URL is provided:

  • Extract the initiative slug from the URL (last path segment before any trailing slash)
  • Call mcp__claude_ai_Linear__get_initiative to verify it exists
  • Call mcp__claude_ai_Linear__save_initiative to update its description with the roadmap content

If LINEAR_URL is empty (create new):

  • Call mcp__claude_ai_Linear__save_initiative with name = roadmap title (from first H1) and description = roadmap content
  • Note the returned initiative URL as LINEAR_URL

5b: Upload research artifacts as initiative documents

Important: mcp__claude_ai_Linear__create_document does NOT support initiativeId. Use the browser-first pattern below for each research file.

Collect research files to upload:

  • {COMMIT_DIR}/research/*.md (all files copied in Step 4)

For each research file:

  1. Browser navigate to the initiative overview page (LINEAR_URL + /overview if not already there)
  2. Browser click "Add document or link…" button (use find tool with query "Add document or link")
  3. Browser click "Create new document…" from the dropdown

- Wait for the tab URL to change to /document/untitled-{slug} — this confirms the blank doc is linked to the initiative - If URL does not change within ~3 seconds, retry the click once. If still no change, log the failure and skip this file.

  1. Extract slug from the tab URL (the part after /document/)
  2. Immediately call mcp__claude_ai_Linear__update_document:

- id: extracted slug - title: humanised filename (e.g. marketResearch: Market Validation) - content: file content — strip all local markdown links before uploading: replace [text](./path), [text](../path), and [text](relative/path) with just text (keep the link text, discard the URL). Absolute https:// links are safe to keep. - Do this BEFORE navigating away — blank docs are garbage-collected on navigation

  1. JS-navigate back to initiative: execute window.location.href = '{LINEAR_URL}/overview' via browser javascript tool (bypasses "Leave site?" dialog)
  2. Repeat for next file

Report each uploaded document title or failure.

Step 6: Notify

if [ "$SILENT" != "true" ]; then
  npx @codevoyant/agent-kit notify \
    --title "pm:approve complete" \
    --message "Roadmap committed to {COMMIT_DIR}"
fi

Report: "Done. Roadmap is at {COMMIT_DIR}/roadmap.md. Research at {COMMIT_DIR}/research/."


PRD Flow

Step 1P: Locate draft

If SLUG provided, resolve to .codevoyant/prds/{SLUG}/{SLUG}.md.

If no SLUG, list directories in .codevoyant/prds/ sorted by modification time and ask:

AskUserQuestion:
  question: "Which draft PRD do you want to approve?"
  header: "Draft PRD"
  options:
    - label: "Most recent draft"
    - label: "I'll specify the slug"

Read the selected PRD. Set DRAFT_PATH and SLUG.

Step 2P: Skip review

PRDs use the pm:prd quality checkpoint inline (Step 4 of pm:prd). No separate review step needed.

Step 3P: Confirm promotion

COMMIT_DIR="docs/prd/${SLUG}"

Check if COMMIT_DIR already exists. If it does, warn:

AskUserQuestion:
  question: "A committed PRD already exists at {COMMIT_DIR}. Overwrite?"
  header: "Overwrite?"
  options:
    - label: "Yes — overwrite"
    - label: "Save as new version (add -v2 suffix to dir name)"
    - label: "Cancel"

Ask for final confirmation:

AskUserQuestion:
  question: "Promote PRD draft to {COMMIT_DIR}?"
  header: "Confirm promotion"
  options:
    - label: "Promote"
    - label: "Cancel"

Step 4P: Promote

mkdir -p "docs/prd/${SLUG}/research"
cp ".codevoyant/prds/${SLUG}/${SLUG}.md" "docs/prd/${SLUG}/${SLUG}.md"
# Copy research from explore dir if present
if [ -d ".codevoyant/explore/${SLUG}" ]; then
  cp ".codevoyant/explore/${SLUG}/summary.md" "docs/prd/${SLUG}/research/" 2>/dev/null || true
  cp ".codevoyant/explore/${SLUG}/research/"*.md "docs/prd/${SLUG}/research/" 2>/dev/null || true
fi

Update agent-kit status:

npx @codevoyant/agent-kit plans update-status --name "${SLUG}-prd" --status Approved

Report: "PRD promoted to docs/prd/{SLUG}/{SLUG}.md."

Step 5P: Linear sync (optional)

If --push flag not passed, ask:

AskUserQuestion:
  question: "Sync this PRD to Linear?"
  header: "Linear sync"
  options:
    - label: "Yes — attach to an existing initiative (I'll provide the URL)"
    - label: "No — skip Linear sync"

If "attach to an existing initiative", ask:

AskUserQuestion:
  question: "Paste the Linear initiative URL:"
  header: "Initiative URL"
  freeform: true

Set LINEAR_URL to the provided value.

If not syncing, skip to Step 6P.

5Pa: Upload PRD as initiative document

Upload the PRD as a Linear document using the browser-first pattern:

  1. Browser navigate to the initiative overview page (LINEAR_URL + /overview if not already there)
  2. Browser click "Add document or link..." button (use find tool with query "Add document or link")
  3. Browser click "Create new document..." from the dropdown

- Wait for the tab URL to change to /document/untitled-{slug} — this confirms the blank doc is linked to the initiative - If URL does not change within ~3 seconds, retry the click once. If still no change, log the failure and skip.

  1. Extract slug from the tab URL (the part after /document/)
  2. Immediately call mcp__claude_ai_Linear__update_document:

- id: extracted slug - title: humanised feature name from SLUG (e.g. slug "persistent-memory" -> "Persistent Memory") — never a generic title like "PRD" or "prd.md" - content: PRD file content — strip all local markdown links before uploading: replace [text](./path), [text](../path), and [text](relative/path) with just text (keep the link text, discard the URL). Absolute https:// links are safe to keep. - Do this BEFORE navigating away — blank docs are garbage-collected on navigation

  1. JS-navigate back to initiative: execute window.location.href = '{LINEAR_URL}/overview' via browser javascript tool

Report uploaded document title or failure.

Step 6P: Notify

if [ "$SILENT" != "true" ]; then
  npx @codevoyant/agent-kit notify \
    --title "pm:approve complete" \
    --message "PRD committed to docs/prd/${SLUG}/${SLUG}.md"
fi

Report: "Done. PRD is at docs/prd/{SLUG}/{SLUG}.md. Research at docs/prd/{SLUG}/research/."

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

34.75%
按下载量换算28

Claude

30.23%
按下载量换算24

Cursor

20.31%
按下载量换算16

Gemini CLI

8.83%
按下载量换算7

安全审计

暂无安全审计结果可展示。

权限和风险

操作浏览器

该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills