AGENT SKILL

create-plans

Create hierarchical project plans optimized for solo agentic development. Use when planning projects, phases, or tasks that Claude will execute. Produces Claude-executable plans with verification criteria, not enterprise documentation. Handles briefs, roadmaps, phase plans, and context handoffs.

License 许可
MIT
最近更新
2026-06-23
Stars
★ 0
Forks
0

01 前言 Front Matter

name
create-plans
description
Create hierarchical project plans optimized for solo agentic development. Use when planning projects, phases, or tasks that Claude will execute. Produces Claude-executable plans with verification criteria, not enterprise documentation. Handles briefs, roadmaps, phase plans, and context handoffs.

02 资源文件 31

03 SKILL.md 正文

<essential_principles>

<principle name="solo_developer_plus_claude"> You are planning for ONE person (the user) and ONE implementer (Claude). No teams. No stakeholders. No ceremonies. No coordination overhead. The user is the visionary/product owner. Claude is the builder. </principle>

<principle name="plans_are_prompts"> PLAN.md is not a document that gets transformed into a prompt. PLAN.md IS the prompt. It contains:

  • Objective (what and why)
  • Context (@file references)
  • Tasks (type, files, action, verify, done, checkpoints)
  • Verification (overall checks)
  • Success criteria (measurable)
  • Output (SUMMARY.md specification)

When planning a phase, you are writing the prompt that will execute it. </principle>

<principle name="scope_control"> Plans must complete within ~50% of context usage to maintain consistent quality.

The quality degradation curve:

  • 0-30% context: Peak quality (comprehensive, thorough, no anxiety)
  • 30-50% context: Good quality (engaged, manageable pressure)
  • 50-70% context: Degrading quality (efficiency mode, compression)
  • 70%+ context: Poor quality (self-lobotomization, rushed work)

Critical insight: Claude doesn't degrade at 80% - it degrades at ~40-50% when it sees context mounting and enters "completion mode." By 80%, quality has already crashed.

Solution: Aggressive atomicity - split phases into many small, focused plans.

Examples:

  • 01-01-PLAN.md - Phase 1, Plan 1 (2-3 tasks: database schema only)
  • 01-02-PLAN.md - Phase 1, Plan 2 (2-3 tasks: database client setup)
  • 01-03-PLAN.md - Phase 1, Plan 3 (2-3 tasks: API routes)
  • 01-04-PLAN.md - Phase 1, Plan 4 (2-3 tasks: UI components)

Each plan is independently executable, verifiable, and scoped to 2-3 tasks maximum.

Atomic task principle: Better to have 10 small, high-quality plans than 3 large, degraded plans. Each commit should be surgical, focused, and maintainable.

Autonomous execution: Plans without checkpoints execute via subagent with fresh context - impossible to degrade.

See: references/scope-estimation.md </principle>

<principle name="human_checkpoints"> Claude automates everything that has a CLI or API. Checkpoints are for verification and decisions, not manual work.

Checkpoint types:

  • checkpoint:human-verify - Human confirms Claude's automated work (visual checks, UI verification)
  • checkpoint:decision - Human makes implementation choice (auth provider, architecture)

Rarely needed: checkpoint:human-action - Only for actions with no CLI/API (email verification links, account approvals requiring web login with 2FA)

Critical rule: If Claude CAN do it via CLI/API/tool, Claude MUST do it. Never ask human to:

  • Deploy to Vercel/Railway/Fly (use CLI)
  • Create Stripe webhooks (use CLI/API)
  • Run builds/tests (use Bash)
  • Write .env files (use Write tool)
  • Create database resources (use provider CLI)

Protocol: Claude automates work → reaches checkpoint:human-verify → presents what was done → waits for confirmation → resumes

See: references/checkpoints.md, references/cli-automation.md </principle>

<principle name="deviation_rules"> Plans are guides, not straitjackets. Real development always involves discoveries.

During execution, deviations are handled automatically via 5 embedded rules:

  1. Auto-fix bugs - Broken behavior → fix immediately, document in Summary
  2. Auto-add missing critical - Security/correctness gaps → add immediately, document
  3. Auto-fix blockers - Can't proceed → fix immediately, document
  4. Ask about architectural - Major structural changes → stop and ask user
  5. Log enhancements - Nice-to-haves → auto-log to ISSUES.md, continue

No user intervention needed for Rules 1-3, 5. Only Rule 4 (architectural) requires user decision.

All deviations documented in Summary with: what was found, what rule applied, what was done, commit hash.

Result: Flow never breaks. Bugs get fixed. Scope stays controlled. Complete transparency.

See: workflows/execute-phase.md (deviation_rules section) </principle>

<principle name="ship_fast_iterate_fast"> No enterprise process. No approval gates. No multi-week timelines. Plan → Execute → Ship → Learn → Repeat.

Milestone-driven: Ship v1.0 → mark milestone → plan v1.1 → ship → repeat. Milestones mark shipped versions and enable continuous iteration. </principle>

<principle name="milestone_boundaries"> Milestones mark shipped versions (v1.0, v1.1, v2.0).

Purpose:

  • Historical record in MILESTONES.md (what shipped when)
  • Greenfield → Brownfield transition marker
  • Git tags for releases
  • Clear completion rituals

Default approach: Extend existing roadmap with new phases.

  • v1.0 ships (phases 1-4) → add phases 5-6 for v1.1
  • Continuous phase numbering (01-99)
  • Milestone groupings keep roadmap organized

Archive ONLY for: Separate codebases or complete rewrites (rare).

See: references/milestone-management.md </principle>

<principle name="anti_enterprise_patterns"> NEVER include in plans:

  • Team structures, roles, RACI matrices
  • Stakeholder management, alignment meetings
  • Sprint ceremonies, standups, retros
  • Multi-week estimates, resource allocation
  • Change management, governance processes
  • Documentation for documentation's sake

If it sounds like corporate PM theater, delete it. </principle>

<principle name="context_awareness"> Monitor token usage via system warnings.

At 25% remaining: Mention context getting full At 15% remaining: Pause, offer handoff At 10% remaining: Auto-create handoff, stop

Never start large operations below 15% without user confirmation. </principle>

<principle name="user_gates"> Never charge ahead at critical decision points. Use gates:

  • AskUserQuestion: Structured choices (2-4 options)
  • Inline questions: Simple confirmations
  • Decision gate loop: "Ready, or ask more questions?"

Mandatory gates:

  • Before writing PLAN.md (confirm breakdown)
  • After low-confidence research
  • On verification failures
  • After phase completion with issues
  • Before starting next phase with previous issues

See: references/user-gates.md </principle>

<principle name="git_versioning"> All planning artifacts are version controlled. Commit outcomes, not process.

  • Check for repo on invocation, offer to initialize
  • Commit only at: initialization, phase completion, handoff
  • Intermediate artifacts (PLAN.md, RESEARCH.md, FINDINGS.md) NOT committed separately
  • Git log becomes project history

See: references/git-integration.md </principle>

</essential_principles>

<context_scan> Run on every invocation to understand current state:

# Check git status
git rev-parse --git-dir 2>/dev/null || echo "NO_GIT_REPO"

# Check for planning structure
ls -la .planning/ 2>/dev/null
ls -la .planning/phases/ 2>/dev/null

# Find any continue-here files
find . -name ".continue-here.md" -type f 2>/dev/null

# Check for existing artifacts
[ -f .planning/BRIEF.md ] && echo "BRIEF: exists"
[ -f .planning/ROADMAP.md ] && echo "ROADMAP: exists"

If NO_GIT_REPO detected: Inline question: "No git repo found. Initialize one? (Recommended for version control)" If yes: git init

Present findings before intake question. </context_scan>

<domain_expertise> Optional: load a domain-expertise skill if one is installed.

If you have a separate skill that bundles deep, stack-specific knowledge (conventions, pitfalls, recommended architecture for a particular kind of project), load it before creating a roadmap or phase plan. This is a soft dependency — if no such skill is installed, proceed without it. Nothing here requires a domain skill to exist.

<scan_domains> Check whether any domain-expertise skill is available:

ls ~/.claude/skills/ 2>/dev/null

If no matching domain skill is found: proceed without domain expertise (graceful degradation). This skill works fine without domain-specific context. </scan_domains>

<inference_rules> If the user's request points at a specific stack (a native mobile/desktop app, a game engine, an Agent SDK app, a Python automation, a frontend, etc.) and a matching domain-expertise skill happens to be installed, infer it and confirm:

Detected: [domain] project → matching domain skill: [skill-name]
Load this expertise for planning? (Y / proceed without it)

</inference_rules>

<load_domain> If a domain-expertise skill is installed and selected, load it intelligently:

Step 1: Read the domain SKILL.md

cat ~/.claude/skills/[domain-skill]/SKILL.md 2>/dev/null

This loads core principles and routing guidance (~5k tokens).

Step 2: Determine what references are needed

A well-built domain skill contains a <references_index> section that maps planning contexts to specific references.

Example:

<references_index>
**For database/persistence phases:** references/persistence.md, references/concurrency.md
**For UI/layout phases:** references/ui-layout.md, references/design-guidelines.md
**For system integration:** references/platform-integration.md
**Always useful:** references/conventions.md
</references_index>

Step 3: Load only relevant references

Based on the phase being planned (from ROADMAP), load ONLY the references mentioned for that type of work.

# Example: planning a persistence phase
cat ~/.claude/skills/[domain-skill]/references/persistence.md
cat ~/.claude/skills/[domain-skill]/references/conventions.md

Context efficiency:

  • SKILL.md only: ~5k tokens
  • SKILL.md + selective references: ~8-12k tokens
  • All references (old approach): ~20-27k tokens

Announce: "Loaded [domain] expertise ([X] references for [phase-type])."

If no domain skill is installed: proceed without domain expertise — do not block.

If the domain SKILL.md has no references_index: fall back to loading all its references, with a note about context usage. </load_domain>

<when_to_load> Domain expertise should be loaded BEFORE:

  • Creating roadmap (phases should be domain-appropriate)
  • Planning phases (tasks must be domain-specific)

Domain expertise is NOT needed for:

  • Creating brief (vision is domain-agnostic)
  • Resuming from handoff (context already established)
  • Transition between phases (just updating status) </when_to_load> </domain_expertise>

<intake> Based on scan results, present context-aware options:

If handoff found:

Found handoff: .planning/phases/XX/.continue-here.md
[Summary of state from handoff]

1. Resume from handoff
2. Discard handoff, start fresh
3. Different action

If planning structure exists:

Project: [from BRIEF or directory]
Brief: [exists/missing]
Roadmap: [X phases defined]
Current: [phase status]

What would you like to do?
1. Plan next phase
2. Execute current phase
3. Create handoff (stopping for now)
4. View/update roadmap
5. Something else

If no planning structure:

No planning structure found.

What would you like to do?
1. Start new project (create brief)
2. Create roadmap from existing brief
3. Jump straight to phase planning
4. Get guidance on approach

Wait for response before proceeding. </intake>

<routing>

Response Workflow
"brief", "new project", "start", 1 (no structure) workflows/create-brief.md
"roadmap", "phases", 2 (no structure) workflows/create-roadmap.md
"phase", "plan phase", "next phase", 1 (has structure) workflows/plan-phase.md
"chunk", "next tasks", "what's next" workflows/plan-chunk.md
"execute", "run", "do it", "build it", 2 (has structure) EXIT SKILL → Use /run-plan <path> slash command
"research", "investigate", "unknowns" workflows/research-phase.md
"handoff", "pack up", "stopping", 3 (has structure) workflows/handoff.md
"resume", "continue", 1 (has handoff) workflows/resume.md
"transition", "complete", "done", "next" workflows/transition.md
"milestone", "ship", "v1.0", "release" workflows/complete-milestone.md
"guidance", "help", 4 workflows/get-guidance.md

Critical: Plan execution should NOT invoke this skill. Use /run-plan for context efficiency (skill loads ~20k tokens, /run-plan loads ~5-7k).

After reading the workflow, follow it exactly. </routing>

<hierarchy> The planning hierarchy (each level builds on previous):

BRIEF.md          → Human vision (you read this)
    ↓
ROADMAP.md        → Phase structure (overview)
    ↓
RESEARCH.md       → Research prompt (optional, for unknowns)
    ↓
FINDINGS.md       → Research output (if research done)
    ↓
PLAN.md           → THE PROMPT (Claude executes this)
    ↓
SUMMARY.md        → Outcome (existence = phase complete)

Rules:

  • Roadmap requires Brief (or prompts to create one)
  • Phase plan requires Roadmap (knows phase scope)
  • PLAN.md IS the execution prompt
  • SUMMARY.md existence marks phase complete
  • Each level can look UP for context </hierarchy>

<output_structure> All planning artifacts go in .planning/:

.planning/
├── BRIEF.md                    # Human vision
├── ROADMAP.md                  # Phase structure + tracking
└── phases/
    ├── 01-foundation/
    │   ├── 01-01-PLAN.md       # Plan 1: Database setup
    │   ├── 01-01-SUMMARY.md    # Outcome (exists = done)
    │   ├── 01-02-PLAN.md       # Plan 2: API routes
    │   ├── 01-02-SUMMARY.md
    │   ├── 01-03-PLAN.md       # Plan 3: UI components
    │   └── .continue-here-01-03.md  # Handoff (temporary, if needed)
    └── 02-auth/
        ├── 02-01-RESEARCH.md   # Research prompt (if needed)
        ├── 02-01-FINDINGS.md   # Research output
        ├── 02-02-PLAN.md       # Implementation prompt
        └── 02-02-SUMMARY.md

Naming convention:

  • Plans: {phase}-{plan}-PLAN.md (e.g., 01-03-PLAN.md)
  • Summaries: {phase}-{plan}-SUMMARY.md (e.g., 01-03-SUMMARY.md)
  • Phase folders: {phase}-{name}/ (e.g., 01-foundation/)

Files sort chronologically. Related artifacts (plan + summary) are adjacent. </output_structure>

<reference_index> All in references/:

Structure: directory-structure.md, hierarchy-rules.md Formats: handoff-format.md, plan-format.md Patterns: context-scanning.md, context-management.md Planning: scope-estimation.md, checkpoints.md, milestone-management.md Process: user-gates.md, git-integration.md, research-pitfalls.md Domain: domain-expertise.md (guide for creating context-efficient domain skills) </reference_index>

<templates_index> All in templates/:

Template Purpose
brief.md Project vision document with current state
roadmap.md Phase structure with milestone groupings
phase-prompt.md Executable phase prompt (PLAN.md)
research-prompt.md Research prompt (RESEARCH.md)
summary.md Phase outcome (SUMMARY.md) with deviations
milestone.md Milestone entry for MILESTONES.md
issues.md Deferred enhancements log (ISSUES.md)
continue-here.md Context handoff format
</templates_index>

<workflows_index> All in workflows/:

Workflow Purpose
create-brief.md Create project vision document
create-roadmap.md Define phases from brief
plan-phase.md Create executable phase prompt
execute-phase.md Run phase prompt, create summary
research-phase.md Create and run research prompt
plan-chunk.md Plan immediate next tasks
transition.md Mark phase complete, advance
complete-milestone.md Mark shipped version, create milestone entry
handoff.md Create context handoff for pausing
resume.md Load handoff, restore context
get-guidance.md Help decide planning approach
</workflows_index>

<success_criteria> Planning skill succeeds when:

  • Context scan runs before intake
  • Appropriate workflow selected based on state
  • PLAN.md IS the executable prompt (not separate)
  • Hierarchy is maintained (brief → roadmap → phase)
  • Handoffs preserve full context for resumption
  • Context limits are respected (auto-handoff at 10%)
  • Deviations handled automatically per embedded rules
  • All work (planned and discovered) fully documented
  • Domain expertise loaded intelligently (SKILL.md + selective references, not all files)
  • Plan execution uses /run-plan command (not skill invocation) </success_criteria>

04 同仓库更多 Skills

Elad73
Elad73/agentic-base-layer

wrap-up

Session-end safe closeout. Updates STATUS, verifies you are on a feature branch (never main), security-greps the diff for secrets, makes a conventional commit (no AI-tool mentions), then asks for approval before pushing and opening a PR. Use this to properly close a development session. Side-effecting and gated on explicit user approval; run explicitly.

★ 0
Elad73
Elad73/agentic-base-layer

wake-up

Session-start context sync. Reads project STATUS, git state, and runs a quick health check, then emits a short (<15 line) announcement of where things stand and the single best next step. Use this at the beginning of a working session to re-establish context fast. Does NOT re-read CLAUDE.md (already in the system prompt). Read-only and fast.

★ 0
Elad73
Elad73/agentic-base-layer

station

End-of-session composition that chains retrospect → document → cleanup → wrap-up into one closeout pipeline. Use this to fully close a working session — capture learnings, sync docs, prune artifacts, then commit and open a PR — without forgetting a step. Side-effecting; honors each sub-skill's confirmation and approval gates. Run explicitly.

★ 0
Elad73
Elad73/agentic-base-layer

retrospect

Post-session learning capture. Mines the current conversation for reusable patterns, pitfalls, architecture decisions, performance and security insights, then appends them to the project knowledge registry. Use this at the end of a working session, after a bug is fixed, or after a non-obvious decision is made — so the next session starts smarter. Writes structured entries to .claude/knowledge/REGISTRY.md.

★ 0
Elad73
Elad73/agentic-base-layer

pre-ship

Mandatory pre-commit quality gates — type-check, lint, test, build — run in order and halting on the first failure. Use this BEFORE committing any change that touches application code, so failures are caught while context is still fresh instead of by CI. Surfaces the failing tool's verbatim error and refuses to proceed until gates pass.

★ 0
Elad73
Elad73/agentic-base-layer

document

Code-is-truth documentation sync. Detects drift between the codebase and project docs (STATUS.md, CLAUDE.md, README), then applies targeted edits so docs describe what the code actually is. Use this after shipping features, before a release, or when docs feel stale. Secret-scans every doc before writing. Makes small surgical edits, never wholesale rewrites.

★ 0