Curate golangci-lint linter set beyond the v2 standard set

[grantcarthew]

Background

Project 06-linter-config-and-print-helpers.md lands .golangci.yml at the repo root with linters.default: standard — the upstream-curated five: errcheck, govet, ineffassign, staticcheck, unused. Running linters.default: all against the codebase during project review surfaced 945 issues across 47 linters. That blast radius is too large to triage as part of the print-strip project, so broader linter adoption is deferred to this follow-up.

Goal

Walk the golangci-lint v2 linter catalogue, decide keep/disable per linter, fix the findings of the ones we keep (or scope them as their own sub-issues), and commit the resulting .golangci.yml with linters.default: all plus a linters.disable list.

Scope

In scope:

• Review each of the ~47 non-standard linters surfaced by linters.default: all and make a keep/disable decision per linter, with a one-line rationale in the config (YAML comment).
• For each keeper, either fix the findings inline if mechanical, or open a sub-issue if the change is non-trivial or stylistic enough to warrant deliberate review.
• Update .golangci.yml to linters.default: all with a linters.disable list and any per-linter settings the keepers need.
• Re-run scripts/invoke-linter and confirm zero findings.

Out of scope:

• Changing errcheck's exclude-functions list for the Fprint* family (set by the print-strip project).
• Re-introducing the _, _ = fmt.Fprint* pattern.

Baseline finding counts from linters.default: all

For context — counts are per-linter from a single run; many hit the 50-per-linter cap so actual totals are higher:

Linter	Count	Linter	Count	Linter	Count
cyclop	50	err113	50	exhaustruct	50
funlen	46	gochecknoglobals	27	gocognit	37
goconst	50	gosec	16	lll	50
mnd	34	nestif	33	paralleltest	50
revive	31	testpackage	25	tparallel	50
varnamelen	50	wrapcheck	50	wsl	50
depguard	50	dupl	19	wsl_v5	16
nlreturn	9	nilerr	12	gocritic	8
nonamedreturns	8	exhaustive	7	unparam	5
staticcheck	5	errorlint	4	dupword	4
ireturn	4	dogsled	3	forbidigo	3
funcorder	3	godot	3	intrange	3
noinlineerr	3	perfsprint	3	nilnil	2
forcetypeassert	2	goprintffuncname	2	noctx	2
usetesting	2	embeddedstructfieldcheck	1	makezero	1
modernize	1	wastedassign	1	whitespace	1

Process suggestion

Group linters into three buckets and walk them:

1. Style/format (wsl, lll, varnamelen, nlreturn, godot, whitespace, dupword) — high-volume stylistic choices. Each is a one-shot decision: do we want this style enforced in the codebase? If yes, the fix is mechanical; if no, disable with rationale.
2. Code quality (cyclop, funlen, gocognit, nestif, dupl, goconst, mnd) — complexity thresholds. Decide thresholds the team is comfortable with, configure, then either refactor or disable on a file-by-file basis with //nolint for legitimate hot spots.
3. Correctness/safety (gosec, errorlint, wrapcheck, err113, nilerr, nilnil, noctx, exhaustive, makezero, wastedassign, forcetypeassert, etc.) — these catch real bugs. Strong bias toward keeping these and fixing the findings.

Acceptance Criteria

• .golangci.yml declares linters.default: all with a linters.disable list whose entries each carry a one-line YAML comment explaining the decision.
• golangci-lint run ./... reports zero issues.
• Any non-trivial fixes that came out of this work are tracked as their own sub-issues or PRs.

References

• Print-strip project that lands standard: 06-linter-config-and-print-helpers.md
• golangci-lint v2 linter catalogue: https://golangci-lint.run/usage/linters/