A personal Claude Code skill that stores your curated engineering rules and actively consults them before Claude drafts code. Complements Claude Code's passive auto-memory: memory surfaces facts when relevant, rules shape the approach up front.
Claude Code
$ npx -y skills add kakha13/think-like-me --agent claude-code
Codex
$ npx skills add kakha13/think-like-me -a codex
Cursor
$ npx skills add kakha13/think-like-me -a cursor
Gemini CLI
$ npx skills add kakha13/think-like-me -a gemini-cli
GitHub Copilot
$ npx skills add kakha13/think-like-me -a github-copilot
OpenCode
$ npx skills add kakha13/think-like-me -a opencode
Windsurf
$ npx skills add kakha13/think-like-me -a windsurf
A personal Claude Code skill that stores your curated engineering rules and actively consults them before Claude drafts code. Complements Claude Code's passive auto-memory: memory surfaces facts when relevant, rules shape the approach up front.
When you ask Claude to build, modify, debug, or design a technical feature, the skill:
mobile, backend, ui, general).rules/<domain>.md file(s).When: keywords overlap with the task.If no rule matches, Claude proceeds normally — no forced matches.
| Invocation | When to use | Flow |
|---|---|---|
/learn-this | Right after correcting Claude | Claude analyzes the correction, drafts a rule with Don't pre-filled from the wrong approach, asks for approval. |
/new-rule [description] | Adding a rule from scratch (not tied to a correction) | Claude parses your prose into the structured block, asks if Why: is missing, shows draft. |
| Auto-detect | You correct Claude without typing a command | Claude offers once: "Caught — want me to save this as a rule? (yes to draft, skip to drop)" — never pressures twice. |
All three end the same: draft → approve → dual-write to rules/<category>.md and ~/.claude/projects/<project>/memory/feedback_<slug>.md (with a pointer in MEMORY.md).
### Short searchable title
**When:** domain + feature keywords + specific context
**Do:** the fix, generalized
**Don't:** what not to do (optional)
**Why:** the mechanism that makes the wrong approach fail
**Example:** short snippet (optional)
When / Do / Why are required. Why in particular matters — it lets Claude judge edge cases the rule doesn't literally cover. Full spec in references/rule-format.md.
~/.claude/skills/think-like-me/
├── SKILL.md ← AI-facing entry point (trigger, flow, dual-write)
├── DESIGN.md ← full design spec
├── PLAN.md ← implementation plan
├── docs/README.md ← this file
├── references/rule-format.md ← canonical rule block spec
├── agents/
│ ├── learn-this.md ← /learn-this procedure
│ └── new-rule.md ← /new-rule procedure
└── rules/
├── mobile.md ← iOS / Android / React Native / Expo
├── backend.md ← Laravel / API / server
├── ui.md ← styling / layout / icons
└── general.md ← cross-cutting (starts empty)
Migrated from cross-project entries in your auto-memory:
Mobile (7): NativeWind crash in tab FlatLists · Expo prebuild required for new plugins · Native module rebuild on dev client · EAS --local /var symlink fix · npm --legacy-peer-deps workaround · Google OAuth PKCE + IdToken clash · .env.local override trap
Backend (2): Laravel custom middleware priority · Sanctum v4.3 no isExpired()
UI (1): No emojis — use icon library
General: empty — accumulates via /new-rule and /learn-this
Project-specific memories (e.g. eSIM Access orderNo fix, processing-tab work, Paddle setup) were intentionally not promoted — they stay in project-scoped memory where they belong.
/new-rule [your description including When, Do, Why]/learn-this/new-rule (e.g. "this is a devops rule") — Claude will propose creating rules/devops.md and ask to confirmrules/<domain>.md in any editor; format stays the sameDESIGN.mdPLAN.mdreferences/rule-format.mdagents/learn-this.md, agents/new-rule.mdStars on GitHub
Helps you take a change from a rough idea or bug report through to a tested, reviewed result, with each step handled by a focused skill. Its skills cover pinning down domain terms, designing module interfaces, diagnosing hard bugs, and reviewing changes against your repo's coding standards.
@mattpocockA development methodology for coding agents that makes the agent brainstorm before building, follow plans, run parallel work and finish branches in a disciplined way. You get fewer skipped steps and more consistent results from your coding assistant.
@obraGives your coding agent a planning, testing and review routine to follow on every task. It plans before building, checks changes with tests, has a fresh reviewer look for regressions, and saves repeated wins as reusable skills.
@affaan-mDeveloper tools for working on the React codebase itself. Provides skills for extracting error messages, managing and debugging feature flags across release channels, running Flow type checks, and fixing lint and formatting issues before committing.
@facebookA personal Claude Code skill that stores your curated engineering rules and actively consults them before Claude drafts code. Complements Claude Code's passive auto-memory: memory surfaces facts when relevant, rules shape the approach up front. Install it with the command on its page. It includes think-like-me. Its skills do not fire on their own yet. Free and open source.
Is this plugin yours?
Claim it with GitHubSubmit a pluginPromote it