Think Like Me

What it does

Think Like Me provides an active reasoning layer: a collection of curated engineering rules the agent consults before drafting, modifying, debugging, or designing technical features. Rules live in rules/<domain>.md and are applied by domain (mobile, backend, ui, general). The skill also defines flows to capture new rules and persist them alongside project memory.

When to use it

Invoke this skill on any coding-related task where domain-specific gotchas matter (mobile timers, platform-specific build quirks, security-sensitive edits). It also triggers on explicit commands like /learn-this, learn this, remember this, and /new-rule, or when the agent detects a user correction and offers to save it.

What's included

• Scripts: none in the repo snapshot (has_scripts=false) but procedural guidance for rule capture and dual-write (rule file + project memory) is included.
• References: no separate references folder in the package (has_references=false).
• Instructions: step-by-step policy for when to consult rules, how to apply them, and how to safely save new rules with an explicit user approval gate.

Compatible agents

Inferred compatibility: Claude Code style agents and any assistant that can read/write local skill and memory files. The design assumes a local filesystem-based Claude Code environment.