Best for
- Use when user asks “分析当前项目”, “后续需求推进”, “项目规划”, “下一步做什么”, “roadmap”, “需求拆解”, “排期”, or wants a project health check plus delivery plan.
hua-bang/pulse-agent/.pulse-coder/skills/project-roadmap/SKILL.md
Use it for operations and research tasks; the detail page covers purpose, installation, and practical steps.
Decision brief
Analyze the current Pulse Coder project/repository and turn vague or concrete needs into prioritized requirements, milestones, tasks, validation plans, and execution handoff.
Compatibility matrix
| Platform | Status | Evidence | What to check |
|---|---|---|---|
| Codex | Not declared | No explicit evidence | Portability before use |
| Claude Code | Not declared | No explicit evidence | Portability before use |
| Cursor | Not declared | No explicit evidence | Portability before use |
| Gemini CLI | Not declared | No explicit evidence | Portability before use |
Installation
The source command is displayed only when detected. A safe inspection prompt is always available so your agent can explain every action before execution.
npx skills add https://github.com/hua-bang/pulse-agent --skill ".pulse-coder/skills/project-roadmap"Inspect the Agent Skill "project-roadmap" from https://github.com/hua-bang/pulse-agent/blob/8245c14cdee086f38128a0e4a39ccff702014ab0/.pulse-coder/skills/project-roadmap/SKILL.md at commit 8245c14cdee086f38128a0e4a39ccff702014ab0. List every install step, command, network request, credential, file read/write, external action, and rollback step. Explain whether it fits my task. Do not install or execute anything until I approve.
Workflow
Identify the user’s intent:
Identify the user’s intent:
Run or inspect, as appropriate:
Produce a concise system map:
Assess the project across these dimensions:
Permission review
The documentation includes network, browsing, or remote request actions.
| Tool search/deferred tools | `packages/engine/src/built-in/tool-search-plugin/`, remote tool registration |The documentation asks the agent to read local files, directories, or repositories.
| “直接干” | Create execution tasks, inspect repo, implement only after scope is clear. |The documentation asks the agent to create, modify, or delete local files.
| “直接干” | Create execution tasks, inspect repo, implement only after scope is clear. |The documentation asks the agent to read local files, directories, or repositories.
| Docs/skill-only change | parse/read file, check frontmatter has `name` and `description`; no build required unless loader changed |Evidence record
| Signal | Value | Evidence type | Meaning |
|---|---|---|---|
| Quality score | 95/100 | Computed | Documentation, specificity, maintenance, and trust rules |
| Repository stars | 28 | Source | Repository attention, not individual Skill quality |
| Compatibility | 0 platforms | Source | Declared in the catalog source record |
| Usage guide | automated source guide | Editorial | Generated or reviewed according to the visible evidence level |
Pinned source
Use this skill to analyze the current project/repository and convert product or engineering needs into an actionable delivery plan.
The expected outcome is not just a summary. It must produce:
Use this skill when the user says things like:
Evidence first.
Project claims must come from files, commands, docs, tests, git history, or explicit user input. If not verified, mark as Assumption.
Protect user work.
Always inspect git status --short before proposing or making code changes. Never overwrite uncommitted user changes unless explicitly instructed.
Separate planning from implementation.
Ask only high-leverage questions.
If critical information is missing, ask at most 3 focused questions. Otherwise proceed with stated assumptions.
Make requirements testable.
Every proposed requirement must include acceptance criteria or a measurable success signal.
Prefer small safe increments.
Break work into reviewable PR-sized slices. Avoid giant “rewrite everything” plans unless the repo evidence makes it unavoidable.
Keep one active execution thread.
For multi-step work, use task tracking: list/create/update tasks, keep one main task in_progress, mark blockers clearly.
When operating in this repository, treat the project as a TypeScript pnpm monorepo.
| Path | Role |
|---|---|
packages/engine | Core agent engine: loop, tools, context, plugins, built-in skill/task/tool-search systems. |
packages/cli | Interactive terminal CLI built on pulse-coder-engine. |
packages/cli (src/tools/sandbox) | Sandboxed JavaScript runtime backing the run_js tool. |
packages/plugin-kit (src/memory) | Memory integration/service module (pulse-coder-plugin-kit/memory). |
packages/plugin-kit | Runtime plugin toolkit, including worktree/vault/devtools helpers. |
packages/acp | ACP client/runner integration. |
packages/engine (src/orchestrator) | Multi-agent orchestration module (subpath export pulse-coder-engine/orchestrator). |
packages/agent-teams | Multi-session collaborative agent coordination. |
packages/canvas-cli | Pulse Canvas CLI and bundled canvas skills. |
apps/remote-server | HTTP/webhook runtime for Feishu/Discord and internal agent API. |
apps/canvas-workspace | Electron canvas workbench. |
src/; build output goes to dist/.pnpm run buildpnpm testpnpm run devpnpm startpnpm --filter pulse-coder-engine typecheckpnpm --filter pulse-coder-engine testpnpm --filter pulse-coder-cli testpnpm --filter @pulse-coder/remote-server buildapps/remote-server is the webhook/server runtime. Its key files include:
src/index.tssrc/server.tssrc/core/dispatcher.tssrc/core/agent-runner.tssrc/core/engine-singleton.tssrc/routes/internal.ts.pulse-coder/skills/<skill-name>/SKILL.md with YAML frontmatter.| Need type | Start investigation here |
|---|---|
| Agent loop behavior | packages/engine/src/core/loop.ts, packages/engine/src/context/, packages/engine/src/ai/ |
| Built-in tools | packages/engine/src/tools/ |
| Plugin loading/hooks | packages/engine/src/plugin/, packages/engine/src/built-in/ |
| Skills system | packages/engine/src/built-in/skills-plugin/, .pulse-coder/skills/ |
| Task tracking | packages/engine/src/built-in/task-tracking-plugin/ |
| Tool search/deferred tools | packages/engine/src/built-in/tool-search-plugin/, remote tool registration |
| CLI UX | packages/cli/src/ |
| Remote Discord/Feishu runtime | apps/remote-server/src/core/, apps/remote-server/src/platforms/, apps/remote-server/src/routes/ |
| Internal scheduled runs | apps/remote-server/src/routes/internal.ts, cron-related tools/routes |
| Memory | packages/plugin-kit/src/memory/, apps/remote-server/src/core/memory-integration.ts |
| Worktree/vault binding | packages/plugin-kit/, apps/remote-server/src/core/worktree/ |
| Canvas workbench | apps/canvas-workspace/, docs/canvas-workspace-product/, docs/06-harness-engineering-roadmap.md |
Identify the user’s intent:
| User intent | Deliverable |
|---|---|
| “分析当前项目” | Project map + health diagnosis + risks + immediate next steps. |
| “需求推进” | Requirement decomposition + priority + milestone plan + task board. |
| “做某个功能” | PRD-lite + technical plan + implementation slices + validation plan. |
| “排期/roadmap” | Sequenced roadmap with dependencies, owners/roles, confidence, and risks. |
| “直接干” | Create execution tasks, inspect repo, implement only after scope is clear. |
If the user does not specify a target area, default to repo-wide analysis plus next 2–4 week plan.
Ask up to 3 questions only if needed:
If unanswered, continue with assumptions.
Run or inspect, as appropriate:
pwd, git status --short, git branch --show-current.package.jsonpnpm-workspace.yamlpackage.jsonREADME.md, README-CN.mdAGENTS.md, CLAUDE.mddocs/, architecture/packages/*/srcapps/*/src*.test.ts, *.spec.tsscriptsgit diff --statgit log --oneline -10Do not read huge generated folders such as dist/, node_modules/, release outputs, session stores, or private memory data unless directly relevant.
Produce a concise system map:
skill tool → detailed instructionsMark each claim as:
Fact — verified from repo evidence.Inference — reasoned from evidence.Assumption — not verified.Assess the project across these dimensions:
| Dimension | Checkpoints |
|---|---|
| Product clarity | Who is the user? What job-to-be-done? What is the current wedge? |
| Architecture | Boundaries, dependency direction, plugin extensibility, runtime coupling. |
| Code quality | Type strictness, duplication, naming, local complexity, generated artifacts. |
| Testability | Existing tests, critical paths covered, missing integration/e2e tests. |
| Developer experience | Build scripts, targeted commands, docs, setup friction. |
| Runtime reliability | retries, abort, compaction, tool failure handling, state persistence. |
| Security | secrets handling, internal API protection, path/file operations, webhook verification. |
| Observability | logs, run IDs, latency traces, devtools, Langfuse/plugin hooks. |
| Delivery risk | blockers, unclear requirements, migrations, third-party APIs, platform quirks. |
Use severity:
P0 — blocks correctness/security/release.P1 — important for near-term delivery.P2 — improves maintainability or UX.P3 — future optimization.For each need, create a PRD-lite block:
## Requirement: <name>
- Problem:
- Target user / scenario:
- Desired outcome:
- Non-goals:
- Constraints:
- Acceptance criteria:
1. ...
2. ...
- Success metrics:
- Dependencies:
- Risks:
- Open questions:
Then decompose into engineering slices:
### Slice <N>: <small deliverable>
- Scope:
- Files likely touched:
- Implementation notes:
- Tests:
- Docs/ops updates:
- Rollback plan:
- Definition of Done:
Slices should usually fit in one focused PR/MR.
Use a lightweight scoring model unless the user provides another one.
Score each candidate 1–5:
| Factor | Meaning |
|---|---|
| Impact | User/business value if completed. |
| Urgency | Time sensitivity or blocking nature. |
| Confidence | Evidence quality and requirement clarity. |
| Effort inverse | 5 = small/easy, 1 = large/uncertain. |
| Risk reduction | How much this reduces future delivery risk. |
Suggested priority score:
Priority = Impact + Urgency + Confidence + EffortInverse + RiskReduction
Then classify:
Now / P0: must do immediately.Next / P1: next milestone.Later / P2: valuable but not blocking.Park / P3: explicitly defer.Also identify dependencies: “A must happen before B because …”.
Produce a roadmap at three resolutions:
Each milestone must include:
## Milestone <name>
- Goal:
- User-visible outcome:
- Work slices:
- Validation plan:
- Release/rollback notes:
- Risks and mitigations:
- Exit criteria:
For implementation handoff, create a task plan:
| Task | Priority | Owner/Agent | Dependencies | Validation | Status |
|---|---:|---|---|---|---|
If task-tracking tools are available, create/update tasks for accepted execution work.
Every plan must include validation commands or checks.
For this repo, prefer targeted checks:
| Area | Suggested checks |
|---|---|
| Engine loop/tools/plugins | pnpm --filter pulse-coder-engine test; pnpm --filter pulse-coder-engine typecheck |
| CLI | pnpm --filter pulse-coder-cli test; pnpm --filter pulse-coder-cli build |
| Sandbox | pnpm --filter pulse-coder-cli test (sandbox specs live in src/tools/sandbox) |
| Memory | pnpm --filter pulse-coder-plugin-kit test; pnpm --filter pulse-coder-plugin-kit build |
| Remote server | pnpm --filter @pulse-coder/remote-server build |
| Cross-package change | pnpm run build; targeted tests first, full build last |
| Docs/skill-only change | parse/read file, check frontmatter has name and description; no build required unless loader changed |
If validation cannot run, state why and provide the next best evidence.
Use this structure by default:
# 项目分析与需求推进方案
## 1. 结论先行
- 当前判断:
- 最应该做的 3 件事:
- 最大风险:
## 2. 证据来源
| Evidence | Path/Command | What it supports | Confidence |
|---|---|---|---|
## 3. 当前项目地图
- 产品定位:
- 架构层次:
- 关键模块:
- 主要流程:
## 4. 健康诊断
| Area | Finding | Severity | Evidence | Recommendation |
|---|---|---:|---|---|
## 5. 需求拆解
<PRD-lite blocks>
## 6. 优先级排序
| Candidate | Impact | Urgency | Confidence | EffortInverse | RiskReduction | Priority | Tier |
|---|---:|---:|---:|---:|---:|---:|---|
## 7. Roadmap
### This week
### Next 2–4 weeks
### Later
## 8. 执行任务板
| Task | Priority | Dependencies | Validation | Status |
|---|---:|---|---|---|
## 9. 开放问题与决策点
- Decision needed:
- Assumption to verify:
## 10. 下一步
- Recommended immediate next action:
- If you want me to execute: <specific first slice>
Before finalizing, verify:
git status --short was checked or explicitly unnecessary.This skill is complete when the user receives either:
Frequently asked questions
Analyze the current Pulse Coder project/repository and turn vague or concrete needs into prioritized requirements, milestones, tasks, validation plans, and execution handoff.
The source record exposes this install command: npx skills add https://github.com/hua-bang/pulse-agent --skill ".pulse-coder/skills/project-roadmap". Inspect the command and pinned source before running it.
Static rules flagged network, read-files, write-files in the source; the page lists the matching lines and excerpts.
Alternatives
alirezarezvani/claude-skills
App Store Optimization (ASO) toolkit for researching keywords, analyzing competitor rankings, generating metadata suggestions, and improving app visibility on Apple App Store and Google Play Store. Use when the user asks about ASO, app store rankings, app metadata, app titles and descriptions, app store listings, app visibility, or mobile app marketing on iOS or Android. Supports keyword research and scoring, competitor keyword analysis, metadata optimization, A/B test planning, launch checklist
wanshuiyin/Auto-claude-code-research-in-sleep
Use it for operations and research tasks; the detail page covers purpose, installation, and practical steps.
prowler-cloud/prowler
PostgreSQL indexing best practices for Prowler: index design, partial indexes, partitioned table indexing, EXPLAIN ANALYZE validation, concurrent operations, monitoring, and maintenance. Trigger: When creating or modifying PostgreSQL indexes, analyzing query performance with EXPLAIN, debugging slow queries, reviewing index usage statistics, reindexing, dropping indexes, or working with partitioned table indexes. Also trigger when discussing index strategies, partial indexes, or index maintenance
brucesongs/kali-claw
Insecure Design (OWASP A06:2025) focuses on security flaws in system architecture and design phases, rather than code implementation-level bugs.