docs(r16): The Go toolbox — example 47 + chapter 47 (Appendices: Tooling)

[patterncatalyst]

Third chapter of Part 13 — Appendices: Tooling: the Go toolchain at reference depth, the twin of ch46. Single-language (langs: [go]), local, offline-verifiable.

What's in it

• Example 47-go-toolbox/ — a stdlib-only Go 1.26 module engineered so every go-native tool produces a checkable effect. verify.lua → PASS 16/0 offline.
• Chapter _docs/47-go-toolbox.md (983 lines) — go/toolchain version management, go generate diff-clean discipline, gofmt vs gofumpt, .golangci.yml/staticcheck authoring, govulncheck workflow, delve configs, pprof/benchstat.
• 2 diagrams — 47-go-toolchain-pipeline (toolchain/generate flow) and 47-go-tool-gates (the two verification lanes).

Verified vs shown

The go-native hard core (A–E) is verified offline — everything ships in the go binary:

• deterministic toolbox report digest — 0x481984990deee5ff, byte-identical to ch46's C++ digest (FNV-1a is language-independent — a cross-appendix check).
• go generate ./... && git diff --exit-code clean (a self-contained generator, no stringer).
• gofmt -l, go vet -tags vetdemo (printf at vetdemo.go:13), go tool pprof -top naming hotLoop.

The six marquee tools — gofumpt, golangci-lint, staticcheck, govulncheck, dlv, benchstat — are absent on this host, so per the chosen offline scope they ship with real committed configs (.golangci.yml, .dlvinit) and skip-if-present gates a reader with the tools would hit, but are marked status--unverified. govulncheck is covered as workflow with an explicitly-placeholder advisory (no vulnerable dep committed, no network).

Differentiation

Cross-references — not re-teaches — ch31 (3-language profiling breadth) and ch44 (GMP runtime internals). ch47 is the tooling that measures the runtime.

Gate

• verify.lua PASS 16/0 (offline) · scripts/validate.py → validate: OK · test-all-examples.py --only 47-go-toolbox → 1 passed
• lgtm-relay: Opus plan → Sonnet execute → Opus validate (12/12 MET, verdict SHIP).