What does this skill do, and when should you use it?
This skill positions you as an orchestrator, delegating fact-gathering and extraction to subagents (docs, examples, signatures, greps) while you handle synthesis and plan authoring. It enforces a mandatory reporting contract: every subagent must list sources, concrete findings, copy-ready snippet locations, confidence notes, and known gaps. Plans begin with Phase 0: Documentation Discovery, which identifies actual APIs, methods, and anti-patterns. Each subsequent phase includes what to implement (framed as copying from docs), documentation references, verification checklists, and anti-pattern guards. The final phase verifies implementations against documentation, greps for anti-patterns, and runs tests. The skill is designed to produce a plan that can be executed consecutively in new chat contexts, with each phase self-contained and documented.
The skill instructs the model to create a detailed, phased implementation plan for features, tasks, or multi-step implementations. It mandates deploying 'Documentation Discovery' subagents before planning to search and read relevant docs, examples, and patterns, compile an 'Allowed APIs' list, and note anti-patterns. The orchestrator consolidates findings into a Phase 0 output. Subsequent phases define tasks to copy from docs, cite specific files/lines, include verification checklists, and list anti-pattern guards. The final phase verifies all implementations match documentation, greps for known bad patterns, and runs tests. The skill also enforces a strict subagent reporting contract with sources, findings, snippet locations, and confidence notes.
- A developer planning a complex feature before coding, ensuring the plan is based on actual APIs rather than assumptions.
- An engineer tasked with a large refactor wants a phased plan with documentation references and verification steps for each phase.
- A tech lead breaking down implementation into self-contained phases that can be executed in new chat contexts.
- An agent developer seeking to prevent hallucinated API methods by requiring every plan step to trace back to documentation.
How do you install this skill?
- The skill relies on external docs and subagents; users should verify information sources when using it.
- The skill description lacks explicit capability boundaries, which may cause false triggers.
- No tests or validation examples are provided for this skill; users should validate generated plans themselves.
- Chinese users should note the skill documentation is in English, and mainland network compatibility is not addressed.
- Shell / CLI
- Local filesystem
The skill resides in the openclaw/skills/make-plan/ directory within the claude-mem repository. The repo is a collection of 21 skills and can be installed as a whole, e.g., via npx claude-mem install or the plugin marketplace steps in the README. However, standalone installation instructions for this specific skill are not provided.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/thedotmack/claude-mem.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/openclaw/skills/make-plan" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Use make-plan to create a plan for [feature/task].
Request a plan in a prompt, e.g., 'Use make-plan to create a plan for [feature/task].' The skill will guide you through Phase 0 documentation discovery and then generate a phased implementation plan with documentation references and verification checklists. It is intended to be used before executing with do.
What are this skill's strengths and limitations?
- Forces planning against real APIs via documentation discovery, reducing hallucination.
- Enforces source and evidence reporting from subagents, improving reliability.
- Each phase is self-contained, facilitating execution across sessions.
- Includes anti-pattern guards to prevent use of nonexistent APIs or parameters.
- Exists only as a managerial skill, providing no out-of-box functionality; requires subagent orchestration.
- Reliance on subagents may add complexity and overhead.
- No evidence of a test suite for the skill itself.
- Bundled with other skills in the repository, which may introduce unrelated dependencies.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Make Plan this page | 51 · Use with care | ★ 98k | 3d ago | Apache-2.0 |
| Plan Execution Orchestrator | 51 · Use with care | ★ 98k | 3d ago | Apache-2.0 |
| Executing Plans | 54 · Use with care | ★ 297k | 16d ago | MIT |
| Hermes Ephemeral Delegation | 49 · Use with care | ★ 7.6k | 3d ago | MIT |
| DeepChat Spec-Driven Development Skill | 58 · Recommended | ★ 6.4k | 3d ago | Apache-2.0 |
How did FollowSkills review this skill?
The skill itself does not involve high-risk operations; it mainly instructs subagents for fact-gathering and planning, with no obvious permission abuse or data leakage risks. However, it does not require user confirmation or provide data-flow transparency, and as an open-source skill, source credibility is average, so points deducted.
The skill provides a clear subagent reporting contract and plan structure, but relies on external docs and subagent reliability; edge cases like undefined failure handling and no tests or error handling notes, so points deducted.
The skill applies to planning tasks with clear triggers, but lacks explicit capability boundaries and non-fit scenarios, and does not mention Chinese support or mainland network reachability, so points deducted.
The skill documentation is well-structured with phased guidance and examples, but lacks versioning, update history, maintenance responsibility, and known-limitation disclosure, so points deducted.
The core task is clear and can guide plan generation, but output format is unverified and depends on LLM quality; marginal value is limited and no concrete result examples are provided, so points deducted.
The repo has CI and tests, but specific tests for this skill are not identified; key paths cannot be verified, and docs lack reproducible examples or evidence, so points deducted.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →