mirroring-competitor-codebaseslisted

# Mirroring Competitor Codebases (the Codebase Mirror) ## Overview When a competitor outranks you, the instinct is to read their copy. But a lot of what search and answer engines reward is not written by a writer at all — it is emitted by a template: the JSON-LD block, the heading shape, the internal-link footer, whether the page arrives as HTML or as an empty `<div id="root">`. Those are **codebase decisions**, and they are visible in the HTML they serve. The Codebase Mirror reads that HTML on both sides and answers one question: **what does their template do that yours doesn't?** It reports gaps in one direction only. A pattern *you* ship and they don't is not in the report — this list exists to be acted on, and acting on it means adding what is missing. ## Where the line is with `analyzing-content-gaps` They are complementary and easy to confuse: | | `analyzing-content-gaps` | `mirroring-competitor-codebases` | |---|---|---| | Unit | One query, one page vs the top 3 | Whole site, 3–5 comparable page types per side | | Reads | Prose — entities, sub-questions, depth | Implementation — what the template emits | | Output | `CONTENT_BRIEF.md` — what to write | Ranked gap list — what to build | | Judgment | LLM reads and compares meaning | Deterministic thresholds, no judgment at runtime | Rule of thumb: **"what should this page say?" → content gaps. "what should our template emit?" → this skill.** Run both when a competitor beats you on a query you care about; the fixes