feat(dataset export): add no-strict-asset-verification flag

[dcilke]

Adds a new flag --no-strict-asset-verification to sanity dataset export to disable strict validation of asset hashes.

---

Note

Low Risk
CLI flag plumbing and dependency bump only; default export behavior remains strict verification unless the flag is set.

Overview
Adds --no-strict-asset-verification to sanity dataset export so exports can continue when downloaded assets fail hash or content-length checks. By default, strictAssetVerification stays true and is passed through to @sanity/export (bumped to ^6.2.0).

The command wires the new oclif flag, documents it in examples, and extends tests to assert the option is forwarded to exportDataset.

^{Reviewed by Cursor Bugbot for commit 477c236. Bugbot is set up for automated code reviews on this repo. Configure here.}

[claude]

Claude finished @dcilke's task in 1m 36s —— View job

---

Review

• Read PR diff
• Review export.ts changes
• Review test changes
• Check for changeset
• Post review

---

Issues

1. Missing changeset (required)

This PR adds a new user-facing CLI flag — a runtime behavior change — but no .changeset/*.md file is included. Per repo convention (and visible in prior PRs like pr-1063.md, pr-1114.md), a patch changeset is required.

Add .changeset/<slug>.md:

---
"@sanity/cli": patch
---

Add `--no-strict-asset-verification` flag to `sanity dataset export` to continue exports when an asset fails hash/content-length verification.

Fix this →

2. Flag description has awkward double-negative — packages/@sanity/cli/src/commands/datasets/export.ts:85-86

"Skip aborting the export when an asset fails…" is hard to parse (skip-abort = continue). The parenthetical also doesn't match the style of other flag descriptions in this file (none use parentheticals).

Suggested rewording:

description: "Continue the export when an asset fails hash or content-length verification (logs a warning instead of aborting)",

Or, even tighter and consistent with sibling flags:

description: "Do not abort the export when an asset fails hash or content-length verification",

3. New flag not surfaced in examples — packages/@sanity/cli/src/commands/datasets/export.ts:37-54

Other notable flags (--no-assets, --raw, --types) each have an example entry. Consider adding one for --no-strict-asset-verification so users discover it via sanity dataset export --help. Minor / optional.

---

Notes (no action needed)

• Flag logic (strictAssetVerification: !flags['no-strict-asset-verification']) is correct given oclif's behavior with explicitly-named no-* flags, and matches the existing pattern for no-assets / no-compress / no-drafts.
• Test at line 469-481 covers the --no-strict-asset-verification case; the default-case assertion at line 179 (strictAssetVerification: true) is also good. An additional explicit test that omitting the flag still passes strictAssetVerification: true is already effectively covered by that update.
• @sanity/export bumped from ^6.1.0 → ^6.2.0 in lockfile and package.json — consistent with introducing the new option.
  • Branch: cldx-5406

[github-actions]

📦 Bundle Stats — @sanity/cli

Compared against main (dafed5c3)

@sanity/cli

Metric	Value	vs main (dafed5c)
Internal (raw)	2.1 KB	-
Internal (gzip)	799 B	-
Bundled (raw)	10.97 MB	-
Bundled (gzip)	2.06 MB	-
Import time	850ms	+3ms, +0.4%

bin:sanity

Metric	Value	vs main (dafed5c)
Internal (raw)	1023 B	-
Internal (gzip)	486 B	-
Bundled (raw)	9.84 MB	-
Bundled (gzip)	1.77 MB	-
Import time	1.98s	+11ms, +0.6%

🗺️ View treemap · Artifacts

Details

• Import time regressions over 10% are flagged with ⚠️
• Sizes shown as raw / gzip 🗜️. Internal bytes = own code only. Total bytes = with all dependencies. Import time = Node.js cold-start median.

📦 Bundle Stats — @sanity/cli-core

Compared against main (dafed5c3)

Metric	Value	vs main (dafed5c)
Internal (raw)	96.3 KB	-
Internal (gzip)	22.7 KB	-
Bundled (raw)	21.61 MB	-
Bundled (gzip)	3.42 MB	-
Import time	791ms	-13ms, -1.6%

🗺️ View treemap · Artifacts

Details

• Import time regressions over 10% are flagged with ⚠️
• Sizes shown as raw / gzip 🗜️. Internal bytes = own code only. Total bytes = with all dependencies. Import time = Node.js cold-start median.

📦 Bundle Stats — create-sanity

Compared against main (dafed5c3)

Metric	Value	vs main (dafed5c)
Internal (raw)	908 B	-
Internal (gzip)	483 B	-
Bundled (raw)	931 B	-
Bundled (gzip)	491 B	-
Import time	❌ ChildProcess denied: node	-

Details

• Import time regressions over 10% are flagged with ⚠️
• Sizes shown as raw / gzip 🗜️. Internal bytes = own code only. Total bytes = with all dependencies. Import time = Node.js cold-start median.

[github-actions]

Coverage Delta

File	Statements
packages/@sanity/cli/src/commands/datasets/export.ts	93.5% (±0%)

Comparing 1 changed file against main @ dafed5c321960164da255ff024dda49667cc5ec3

Overall Coverage

Metric	Coverage
Statements	84.3% (±0%)
Branches	74.3% (±0%)
Functions	84.2% (±0%)
Lines	84.8% (±0%)

[runeb]

@claude address the 3 review findings in a new commit

[squiggler-app]

Claude finished @runeb's task in 2m 5s —— View job

---

Addressed review findings in 477c236

• Finding 1 — Missing changeset: Already addressed by auto-generated .changeset/pr-1135.md.
• Finding 2 — Double-negative flag description: Reworded to "Do not abort the export when an asset fails hash or content-length verification" (matches the second tighter suggestion).
• Finding 3 — Surface flag in examples: Added a new entry under examples demonstrating --no-strict-asset-verification.

Diff

+    {
+      command:
+        '<%= config.bin %> <%= command.id %> moviedb moviedb.tar.gz --no-strict-asset-verification',
+      description: 'Export dataset without aborting on asset verification failures',
+    },
...
     'no-strict-asset-verification': Flags.boolean({
       default: false,
       description:
-        'Skip aborting the export when an asset fails hash/content-length verification (keeps the asset with a warning)',
+        'Do not abort the export when an asset fails hash or content-length verification',
     }),

Notes

• I was unable to run pnpm check:lint / pnpm test from this environment (commands required approval that isn't granted to this action). The change is a string/example update with no logic impact, but CI should be the source of truth here — please re-run if you'd like me to address any failures.
  · Branch: cldx-5406

[runeb]

LGTM