spec (1)

1-1: Consider documenting spec changes in the PR or commit.

For maintainability, it would be helpful to document what changed in the spec submodule update, either in the PR description or a commit message. Key changes to document:

• Addition of include_thumbnails query parameter
• Addition of fill_gaps to shot segmentation configuration
• Change in maximum duration bounds (3600 → 600)

This helps reviewers and future maintainers understand the impact of the spec update without needing to diff the submodule commits directly.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@spec` at line 1, Add a short changelog entry to the PR description or the
commit message summarizing the spec submodule update: list the new query
parameter include_thumbnails, the new shot segmentation option fill_gaps, and
the change to maximum duration bounds from 3600 to 600; mention any affected
endpoints or config objects (e.g., shot segmentation configuration) and the
expected behavioral impact so reviewers and future maintainers can quickly
understand the update without diffing the submodule.

src/api/describe.api.ts (1)

80-102: Consider: waitForReady doesn't forward include_thumbnails to final response.

The waitForReady method only forwards response_format when calling getDescribe (line 94). If callers want thumbnails in the final returned job object, they would need to call getDescribe separately after waitForReady completes.

This is likely acceptable since polling typically only checks status, but you may want to consider supporting include_thumbnails in WaitForReadyOptions for consistency if users expect thumbnails in the final response.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@src/api/describe.api.ts` around lines 80 - 102, waitForReady currently passes
only response_format to getDescribe, so callers cannot request thumbnails in the
final returned job; update the WaitForReadyOptions type to include
include_thumbnails (boolean) and forward include_thumbnails when calling
getDescribe inside waitForReady (in addition to response_format) so the final
returned job can contain thumbnails when requested; make sure to read
include_thumbnails from options and pass it through in the getDescribe call.

generate.js (2)

137-140: Document the regex limitation for nested parentheses.

The regex pattern [^)]* cannot handle default values containing nested parentheses (e.g., .default(someFn())). While OpenAPI defaults are typically JSON literals that don't contain nested parens, this assumption could break with unusual specs.

Consider adding a comment documenting this limitation, or for robustness, implementing balanced-parenthesis matching:

📝 Suggested documentation comment

 // Pattern 1: .nullish().default(VALUE) → .nullish()
+// Note: [^)]* assumes no nested parentheses in default values (safe for JSON literals)
 fileContent = fileContent.replace(/\.nullish\(\)\.default\([^)]*\)/g, '.nullish()');

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@generate.js` around lines 137 - 140, The two regex replacements on
fileContent (the patterns using /\.nullish\(\)\.default\([^)]*\)/ and
/\.optional\(\)\.default\([^)]*\)/) do not handle default(...) values that
contain nested parentheses; update generate.js to either (a) add a clear comment
above these replacements documenting this limitation (that the regex won't match
defaults with nested parentheses like .default(someFn())) or (b) replace the
simple regex with a more robust parser/approach that matches balanced
parentheses for the .default(...) portion; reference the fileContent variable
and the two regex replacement lines when making the change.

---

132-159: Consider combining the file read/write with the existing loop.

The new transformation reads and writes each file (lines 132-159), then the existing nullish/nullable fix loop reads and writes the same files again (lines 161-337). This results in two disk I/O cycles per file.

For better efficiency, consider combining into a single loop that applies all transformations before writing once. This is a minor optimization since it's a build-time script.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@generate.js` around lines 132 - 159, The current generatedFiles for-loop
reads each file, applies three .default()/ .optional()/.nullish() regex
transforms and writes it back using fs.readFileSync/fs.writeFileSync, but a
separate later nullish/nullable fix loop re-reads/writes the same files causing
duplicate I/O; consolidate by moving the nullish/nullable fix logic into this
same for (const file of generatedFiles) loop (use the same fileContent variable
and apply all regex replaces — including the existing
.nullish().default/.optional().default/.default(...) handler and the subsequent
nullish/nullable fixes) and call fs.writeFileSync exactly once per file to write
the final fileContent.