dev-pm-flow

dev-pm-flow is a skill for Claude Code, Codex from xl188/XLSkills. It costs 46 tokens per session (3,664 once invoked), scanned A, original, MIT.

A Chinese-language development workflow covering requirements, planning, user confirmation, implementation, review, and delivery.

In plain words
What is it for?
Use it to break requirements into independently verifiable tasks, inspect the existing code, plan changes, implement them, and perform a final review.
Why use it?
It structures development requests so the intended result is confirmed before work begins and the finished changes are checked afterward.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/xl188/xlskills/dev-pm-flow
Any agent
npx skills add xl188/XLSkills --skill dev-pm-flow
Clone the repo
git clone --depth 1 https://github.com/xl188/XLSkills

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for dev-pm-flow

README.md
[![agentmods](https://agentmods.dev/badge/skills/xl188/xlskills/dev-pm-flow.svg)](https://agentmods.dev/skills/xl188/xlskills/dev-pm-flow)
Your own site
<a href="https://agentmods.dev/skills/xl188/xlskills/dev-pm-flow"><img src="https://agentmods.dev/badge/skills/xl188/xlskills/dev-pm-flow.svg" alt="Measured on agentmods" height="20"></a>
Per session 46 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 3,664 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00046 $0.03664
Opus 5 $0.00023 $0.01832
Sonnet 5 $0.00009 $0.00733
Haiku 4.5 $0.00005 $0.00366

Measured 5d ago against content hash 326bf19c0b1f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

dev-pm-flow scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

- **执行中遇难缠 bug → 先建复现闭环**(联动 `systematic-debugging`):失败测试 / curl / 最小复现,要求秒级、确定性、能抓住这个具体 bug;**无闭环不假设**,禁止边猜边改。修前先写回归测试;没有正确的测试接缝时,"没有接缝"本身记入交付说明
dev-pm-flow/SKILL.md · 178 lines

How it starts

The opening of the file, as written. The whole thing — 178 lines — stays where its author put it; the contents beside it link to each section on GitHub.

开发 PM 流程

用户提开发需求时,按此流程走。我是 PM 兼执行者,用户是唯一的方案拍板人和最终验收人。

流程总览

① 需求理解 → ② 拆解+方案 → ③ 用户确认 → ④ 执行开发 → ⑤ 自审 → ⑥ 交付过目 → ⑦ 发布/打回

任何一步卡住都可以向上回退。方案未确认,绝不动手。


① 需求理解

  • 复述需求,确认我理解对了(一句话说清"做什么、为什么")
  • 有歧义直接问,不猜

意图确认(需求含糊 / 非平凡时,先确认意图再进 ②)

用户给的往往是"表面需求"("弄个仪表盘""优化一下"),不是真正想要的。需求含糊、或改动非平凡(多模块 / 跨项目 / 30 分钟以上)时,先用轻量采访把意图锁死:

  1. 说出当前理解 + 置信度:一句话重述,附 0-100% 置信度;<70% 时补一句"缺什么"(让用户知道要补哪块)
  2. 关键点一次一问,不批量——用户对假设的反应比对列表快;第三个问题往往依赖第一个的答案
  3. 拆穿"惯例式回答":用户答得官方/套路("要可扩展""标准做法""干净架构")时,追问一句:"如果不用向任何人交代,你真正想要什么?"——这一句常比前面五句有用
  4. 用用户的话重述,含六要素:做什么 / 给谁 / 为什么现在 / 怎么算成 / 硬约束 / 明确不做("不做"占错配的一半,必须写)
  5. 确认要显式"是":"随便你""听起来不错""走吧"都不算确认(那是授权或礼貌退场),需再问"有要修正的吗?"

意图锁死后再进 ② 拆解+方案,方案里的需求原文 = 意图确认的结论(供 Spec 轴对照)。

② 拆解 + 方案

  • 读相关代码 / 配置,摸清现状
  • 拆成具体任务步骤(改哪些文件、加什么逻辑、什么顺序)
  • 垂直切片原则:每张任务票 = 一条贯穿所有层、可独立验证的端到端切片(能单独演示/验收),不是只切一层的水平切片;每张票带验收标准阻塞关系(blocked by 哪些票;无阻塞的票可立即开工)
  • 大重构走 expand-contract(波及全库的机械改动,如改列名/共享类型/迁移):先加新形态与旧形态并存 → 按爆炸半径分批迁移(每批一张票,保持可编译)→ 最后删旧形态。禁止一把梭
  • 任务分级
    • 🟢 轻量(单文件 / <10 行 / 配置改动)→ 我直接 patch,不走子代理
    • 🟡 中等(2-3 文件 / 新增函数或接口)→ 我自己写,派子代理审查
    • 🔴 重型(多文件重构 / 新模块 / 复杂算法)→ 子代理并行开发,或建议上编码 agent
  • 方案输出格式:目标 + 改动清单 + 涉及文件 + 风险点 + 预计改动量
  • 跨项目影响检查:后端改了 API → 前端是否要跟着调?数据库改了 → 其他服务是否依赖?方案里必须列出所有受影响的项目/模块
  • 数据库变更检查:涉及实体/表结构改动时,方案必须包含 migration 脚本或 SQL 变更语句,不能只改代码不给出库脚本
  • 复杂方案写入 <项目根>/.agent/plans/ 存档(见 ③,直接写文件,不走其他技能)

③ 用户确认

  • 方案给用户,等"可以"/"动手"/"确认"
  • 用户说改 → 回 ② 调方案
  • 没说确认就不动手,可以补充分析但不出代码

方案存档(防上下文丢失)

确认后、执行前,把最终方案写入项目根目录下的 .agent/plans/

  • 路径 = <项目根>/.agent/plans/YYYY-MM-DD_HHMMSS-<需求slug>.md,如 <your-project>/.agent/plans/...
  • 目录不存在则创建mkdir -p .agent/plans);用户主目录(~/.agent)不是存档位置
  • 多项目混用统一工作区(如 <workspace-root>)时,若各项目无独立 .agent,可放工作区根,但必须在方案里写清绝对路径,防止 Spec 轴找错
  • 内容含:需求原文、改动清单、涉及文件、用户确认的要点。即使会话被压缩或跨天,从文件就能恢复上下文继续执行

④ 执行开发

执行前 Checkpoint

动手改代码前,先创建回滚点:

前置检查:确认工作区只含本次任务的改动。 git status 若发现与本次任务无关的未提交改动(用户或其他任务留下的),先跟用户确认怎么处理(单独 stash / 先提交),不把别人的改动混进 checkpoint——否则 pop 时新旧混杂,回滚点失效。

Read the full file on GitHub · 178 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 5d ago First seen · 178 lines · 46 tokens per session scan A 326bf19c0b1f

Subscribe to this mod's changes

dev-pm-flow is a skill published in the GitHub repository xl188/XLSkills (2 stars, last pushed 7d ago), licensed MIT. It adds 46 tokens to every session and 3,664 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

anysearch

Real-time search engine supporting web search, vertical domain search, parallel batch search, and URL content extraction.

anysearch-ai/anysearch-skill · 24 tokens

hermes-self-evaluation

Use this skill when the user asks to audit, review, or optimize Hermes's own performance — analyzing session data, skills, configuration, costs, and usage patterns to identify improvements, automation opportunities, and system optimizations.

AtlasOmnia/donna-starter · 69 tokens

SoloFlow

A meta-skill that silently watches your workflows and automatically generates reusable Hermes skills from them.

SonicBotMan/SoloFlow · 21 tokens

mnemosyne-maintenance

Use when: upgrading Mnemosyne, diagnosing slow/hung consolidation (mnemosynesleep), fixing missing embeddings, or troubleshooting import/version mismatches.

AtlasOmnia/donna-starter · 40 tokens

svix-sending-webhooks

Everything for working with Svix webhooks: first-time setup (API key, SDK install, first message), Dispatch (sending webhooks to your customers), Ingest (receiving third-party webhooks), Applications, Channels, customer UIDs, idempotency, App Portal embedding, operational webhooks, the Svix CLI, and — only when the…

svix/ai · 138 tokens

git-phase-restore

Autonomous Git-based project phase restoration. Uses Git history (commits, tags, branches, diffs, semantic messages) to identify and restore any development phase automatically. Trigger for: "restore to when X worked", "go back to before Y broke", "show project phases", "undo the last feature", "roll back to [phase]"…

mahmoud20138/Tradecraft · 165 tokens