feat!: adapt core and storage clients to the FLAME Hub 0.13.0 API

[tada5hi]

Adapts CoreClient and StorageClient to FLAME Hub 0.13.0, which landed two breaking changes together:

• PrivateAIM/hub#1806 — refactor!: camelCase entity properties, domain types & HTTP API
• PrivateAIM/hub#1801 — feat!: record data/meta envelope, meta.schema discovery and dependency bump

Stacked on #128, whose _unwrap_single_resource hook this builds on. Review/merge #117 → #128 → this.

camelCase

Same treatment AuthClient got in #117: model fields, method keyword arguments and the rapiq filter/sort/fields/include vocabulary all move to camelCase. A pydantic field name is the wire key, so renaming the field is the whole change — no aliases, no compatibility shim, matching hub, which shipped none either.

# before
core_client.create_node(name="my-node", realm_id=master_realm)
core_client.find_nodes(sort={"by": "created_at"})
node.registry_project.external_name

# after
core_client.create_node(name="my-node", realmId=master_realm)
core_client.find_nodes(sort={"by": "createdAt"})
node.registryProject.externalName

What deliberately stays snake_case

Identifier	Why
node_type, bucket_type, registry_project_type	Python-local disambiguators for a hub field called type — there is no hub field by these names
is_entrypoint	Same, for a hub field called root
base_url, chunk_size, UploadFile.file_name / content_type	Transport-level, never on the wire
all of _auth_flows (client_id, client_secret, grant_type, …)	OAuth2/OIDC protocol surface — hub and authup both keep it snake_case
ErrorResponse.status_code	Already accepts both spellings via AliasChoices

Two filter keys that would have failed silently

delete_analysis_logs and delete_analysis_node_logs built filter[analysis_id] / filter[node_id] as string literals. rapiq drops an unknown filter key rather than rejecting it, so against 0.13.0 these would not have raised — they would have widened both DELETEs to an unfiltered selection. Renamed to filter[analysisId] / filter[nodeId].

Record envelope

Hub wraps records as {"data": …, "meta": …} but keeps a handful of endpoints deliberately flat, so unlike authup the envelope is per-endpoint, not per-client. An unconditional client-level override would have broken the credential routes.

• _create_resource, _get_single_resource and _update_resource take a new envelope keyword, defaulting to True.
• The five flat routes pass envelope=False: get_node_registry_credentials, get/update_node_client_credentials, get/update_analysis_client_credentials. Verified against hub's controllers, which declare Promise<ClientCredentials> / Promise<RegistryCredentials> with no envelope.
• send_analysis_command reads the body itself rather than going through the helpers; POST /analyses/:id/command is enveloped (Promise<EntityRecordResponse<Analysis>>), so it now unwraps too.
• CoreClient and StorageClient override _unwrap_single_resource. All three overrides now delegate to a shared unwrap_enveloped_resource() helper that names the required service version in the ValueError, so an outdated deployment is recognizable from the exception alone. BaseClient still returns the body unchanged.

_delete_resource needed no change — it never reads the body, and hub still answers DELETE with 202. Collection responses were already validated as an envelope by ResourceList, and POST /buckets/:id/upload answers with a collection rather than a record, so it is untouched.

Downstream

PrivateAIM/node-hub-api-adapter pins flame-hub-client==0.4.1 and declares FastAPI response_model=list[Project] against these models, so bumping it propagates the camelCase field names to its own wire shape with no model changes on its side. That in turn is what lets PrivateAIM/node-ui consume @privateaim/core-kit types directly instead of generating them from a checked-in swagger.json.

Testing

pytest -m "not integration" — 80 passed (10 new). ruff check and ruff format --check clean. Integration tests need a live 0.13.0 Hub and have not been run here.

New unit coverage in tests/test_base_client.py: core/storage unwrapping across 200/201/202, rejection of a flat body with the version-named ValueError, the credential routes staying flat, list responses untouched, and send_analysis_command going through the hook.

Not included

The models are missing some fields hub 0.13.0 returns — Node.robotId, Project.clientId / robotId, AnalysisBucketFile.robotId. Pre-existing (pydantic ignores extras), unrelated to this rename, and left for a follow-up. One stale robot_id line was dropped from the README dump, which the Node model never actually emitted.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: aa46d4fc-c109-4926-8b8d-88c5da9c9df2

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}