[Critical] SDK Sync: Add 4 missing resources + ManualConnect

[mack-ship-it]

Critical — New Resources

Adds 4 entirely missing resources and 1 missing sub-resource that exist in the OpenAPI spec and Node SDK but were absent from Python:

• ForwardingRequest — ,
• Secret — Full CRUD ()
• Team — Retrieve, create, encryption key, MLE public keys
• ManagedAccount — List, retrieve, transactions
• Entity ManualConnect — Sub-resource for manual account linking

All resources wired into the Method client.

Part of the SDK sync split — see also: High, Medium, Low PRs.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: e053773d5c

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[chatgpt-codex-connector]

Bump Method-Version before exposing new endpoints

These newly wired forwarding_requests and secrets clients still inherit Resource.__init__'s hard-coded method-version: 2025-07-04, but those endpoints are available under the newer API version. When a caller uses method.forwarding_requests.create(...) or method.secrets.create(...), the SDK sends the older version header and the server can reject the endpoint or return an incompatible schema, so the SDK needs to update/configure the Method-Version before exposing these resources.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Expose MLE keys under teams.mle

The documented SDK shape for MLE key management is method.teams.mle.public_keys.*, but this attaches the resource directly as method.teams.public_keys and never defines teams.mle. Users following the documented Python/Node shape will hit AttributeError before any request is sent; wrap this in a mle subresource and put public_keys under it.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Model the full MLE key create payload

For direct registration the API requires type: 'direct' and contact, and for well-known registration it requires type: 'well_known' plus well_known_endpoint; typing the create opts as only jwk makes the documented payload fail type checking and encourages sending {jwk: ...}, which the API rejects. Include the registration type/contact and optional well-known fields in this TypedDict.

Useful? React with 👍 / 👎.

[mack-ship-it]

Codex Review Assessment

Comment 1 (P1) - API version on forwarding_requests/secrets:
False positive. The method-version header is set globally in Resource.init() at resource.py:150. All resources inherit from Resource and automatically use whatever version is set. The version bump is handled in the Low PR (#76). These endpoints don't need a separate version configuration.

Comment 2 (P2) - Team.public_keys path:
Partially valid concern, but the SDK maps to the actual API path /teams/public_keys which matches the OpenAPI spec. If the docs show a different SDK shape (method.teams.mle.public_keys), that's a docs issue not a code issue.

Comment 3 (P2) - MLE public key create opts missing type/contact/well_known_endpoint:
False positive. Checked the OpenAPI spec MLEPublicKeyCreateRequest - the only required field is jwk (object). There is no type, contact, or well_known_endpoint in the schema. The TypedDict with just jwk is correct.