gts-analyze

Thin wrapper around the workspace binary at tools/gts-analyze/. Source code, full docs, and human-readable usage live there: tools/gts-analyze/README.md.

How to invoke

1. Parse the user's invocation for a gear path (absolute or repo-relative). If absent, ask once.
2. Run from the workspace root:

   cargo run --release -p gts-analyze -- <MODULE_PATH> [flags]
   

   First invocation in a fresh checkout compiles the crate (~30s with syn); subsequent runs are instant.
3. Show the markdown report verbatim — don't summarise away the per-type detail.

Flags

• --format json — machine-readable JSON instead of markdown.
• --out FILE — write report to file instead of stdout.
• --include-tests — include Rust test files (default skips integration tests, benches, *_test(s).rs, and items behind #[test] / #[cfg(test)]).
• --skip-docs — drop every *.md and any file under docs/. Note: type definitions sourced from docs/schemas/*.schema.json disappear too — use only for code-only audits.
• -v / --verbose — expand reference lists to per-line file:line — context. Default is compact (file + count).

Report sections

1. Summary — file counts and totals.
2. GTS Types Defined Here — per defined type: schema source (#[gts_type_schema] / *.schema.json / struct_to_gts_schema!), location, struct name when applicable, and usage breakdown across sdk / main / plugin / doc.
3. GTS Instances Defined Here — per declared instance: macro form, inferred typed struct, base type, location, and usage breakdown.
4. Other GTS IDs Referenced — table of IDs not defined here, plus a collapsible per-ID file breakdown.

Tips

• A GTS ID ending with ~ is a Type, otherwise an Instance (gts-spec §8.1).
• For doc-heavy gears with hundreds of .md references, suggest --skip-docs if the user only cares about code usage.
• cargo run builds on first invocation; if the user is running this often, cargo build --release -p gts-analyze once up front avoids latency.