release: 1.22.0

[stainless-app]

Automated Release PR

1.22.0 (2026-08-20)

Full Changelog: v1.21.0...v1.22.0

Features

• internal/types: support eagerly validating pydantic iterators (04e4187)
• stlc: configurable CI runner and private-production-repo support in workflow templates (48c8386)
• support setting headers via env (30a033a)

Bug Fixes

• client: add missing f-string prefix in file type error message (ab34d42)
• client: preserve hardcoded query params when merging with user params (987fbc7)
• ensure file data are only sent as 1 parameter (a7c0bac)
• use correct field name format for multipart file arrays (c335429)

Performance Improvements

• client: optimize file structure copying in multipart requests (ea9d673)

Chores

• internal: allow the mock server port to be set with STAINLESS_MOCK_PORT (23f1a89)
• internal: more robust bootstrap script (2291709)
• internal: reformat pyproject.toml (acc97d1)
• tests: bump steady to v0.22.1 (02f10ea)

---

This pull request is managed by Stainless's GitHub App.

The semver version number is based on included commit messages. Alternatively, you can manually set the version number in the title of this pull request.

For a better experience, it is recommended to use either rebase-merge or squash-merge when merging this pull request.

🔗 Stainless website
📚 Read the docs
🙋 Reach out for help or questions

[cursor]

PR Summary

Medium Risk
Changes affect core request construction (query merge, multipart file extraction/copying) used by all API calls; behavior fixes could alter edge-case requests but are covered by new tests.

Overview
Release 1.22.0 bumps package metadata, changelog, and the tracked OpenAPI spec URL.

HTTP client behavior adds optional default headers from the LUMAAI_CUSTOM_HEADERS environment variable (newline-separated Name: value lines, merged under explicit default_headers). Request building now merges query strings embedded in the request URL with caller params, with explicit params winning on conflicts.

Multipart / file uploads gain configurable array field naming in extract_files (brackets, repeat, comma, indices), safer dict key removal for array-only path tails, and deepcopy_with_paths instead of full-structure deepcopy_minimal when preparing bodies so extract_files mutations do not corrupt caller data.

Pydantic v2 adds an EagerIterable annotated type that eagerly validates iterable fields and avoids empty model_dump after generator consumption.

CI and dev tooling: Stainless SDK repos use depot-ubuntu-24.04 via startsWith(github.repository, 'stainless-sdks/'), GitHub Actions are pinned to commit SHAs, mock tests use Steady v0.22.1 and optional STAINLESS_MOCK_PORT, and bootstrap treats unset SKIP_BREW more safely.

^{Reviewed by Cursor Bugbot for commit 937bd8e. Bugbot is set up for automated code reviews on this repo. Configure here.}

[stainless-app]

🧪 Testing

To try out this version of the SDK:

pip install 'https://pkg.stainless.com/s/luma_ai-python/23f1a89bccccb84f438e2b8bf91928f5c1c746b8/lumaai-1.21.0-py3-none-any.whl'

Expires at: Sat, 19 Sep 2026 04:57:47 GMT
Updated at: Thu, 20 Aug 2026 04:57:47 GMT

[cursor]

Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.

Bugbot Autofix prepared a fix for the issue found in the latest run.

• ✅ Fixed: Array file paths drop non-lists
  • extract_files now pops leftover path segments only when the value is actually a list, so a single file or other non-list stays in the request body instead of being discarded.

Or push these changes by commenting:

@cursor push b7dac59259

Preview (b7dac59259)

diff --git a/src/lumaai/_utils/_utils.py b/src/lumaai/_utils/_utils.py
--- a/src/lumaai/_utils/_utils.py
+++ b/src/lumaai/_utils/_utils.py
@@ -109,8 +109,10 @@
     if is_dict(obj):
         try:
             # Remove the field if there are no more dict keys in the path,
-            # only "<array>" traversal markers or end.
-            if all(p == "<array>" for p in path[index:]):
+            # only "<array>" traversal markers or end. Only pop leftover
+            # "<array>" segments when the value is actually a list.
+            remaining = path[index:]
+            if not remaining or (all(p == "<array>" for p in remaining) and is_list(obj.get(key))):
                 item = obj.pop(key)
             else:
                 item = obj[key]

diff --git a/tests/test_extract_files.py b/tests/test_extract_files.py
--- a/tests/test_extract_files.py
+++ b/tests/test_extract_files.py
@@ -41,6 +41,20 @@
     assert query == {"title": "hello"}
 
 
+def test_array_path_does_not_drop_non_list() -> None:
+    query = {"files": b"single file", "title": "hello"}
+    assert extract_files(query, paths=[["files", "<array>"]]) == []
+    assert query == {"files": b"single file", "title": "hello"}
+
+    nested = {"foo": {"files": b"single file"}, "keep": 1}
+    assert extract_files(nested, paths=[["foo", "files", "<array>"]]) == []
+    assert nested == {"foo": {"files": b"single file"}, "keep": 1}
+
+    query2 = {"files": {"not": "a list"}}
+    assert extract_files(query2, paths=[["files", "<array>"]]) == []
+    assert query2 == {"files": {"not": "a list"}}
+
+
 @pytest.mark.parametrize(
     "query,paths,expected",
     [

_{You can send follow-ups to the cloud agent here.}

Want higher recall? High effort reviews run extra passes and find more bugs. A team admin can switch effort levels in the Cursor dashboard.

^{Reviewed by Cursor Bugbot for commit 937bd8e. Configure here.}

[cursor]

Array file paths drop non-lists

Medium Severity

extract_files now pops a field whenever the remaining path is only <array> markers, before confirming the value is a list. A single file (or other non-list) is removed from the body and never emitted as a multipart part, so that input is dropped from the request entirely.

 

^{Reviewed by Cursor Bugbot for commit 937bd8e. Configure here.}