What problem does it solve?
Understanding how code composes — which modules a request touches, what types cross boundaries, and in what order calls happen — usually requires reading many files and holding the result in memory. This Skill produces a trustworthy, read-only render of that shape where every name, type, and path is cited by file and line or marked [NEW], so the output can be checked against the source instead of trusted from memory.
Core Features & Use Cases
- Cited shape rendering: Produces types crossing boundaries, a module boundary table stating what each module hides, and signatures grouped by module, each verified against source.
- Annotated call graphs: Draws one indented call graph per wiring (production, tests, feature flags) with annotations for conditions, loops, async, errors, and boundary crossings.
- Pre-implementation pseudocode: Renders a [NEW] shape for a planned change, saved beside the slice plan so implementation and tests can be checked against it.
- Use Case: When onboarding to an unfamiliar checkout flow, ask for a render of the entry point; you receive the boundary table, signatures, and a call graph showing exactly which network, database, and async edges the request touches.
Quick Start
Use render-code-shape to render how the checkout request composes, including entry points, boundary types, signatures, and the cited call graph, without changing any code.
