Why

Token activity is useful account-level context, but terminal clients need a supported app-server path to fetch it without reaching into ChatGPT backend details directly. The API should also live under the broader account usage umbrella so future usage surfaces can be added without proliferating user-facing concepts.

What Changed

• Add codex-backend-client support for the ChatGPT profile token-usage payload.
• Add the v2 account/usage/read app-server RPC.
• Map lifetime usage, peak daily usage, streak, longest task duration, and daily buckets into app-server protocol types.
• Gate the request on Codex-backend auth, which supports ChatGPT auth tokens and AgentIdentity.
• Regenerate the app-server JSON and TypeScript schema fixtures.

Token Count Source

account/usage/read returns the token-usage aggregate supplied by the ChatGPT profile backend. App-server maps that backend-owned aggregate into protocol fields; it does not recompute cached-token treatment, usage multipliers, or raw input/output totals locally.

Stack

1. feat(app-server): expose account token usage [1 of 2] (this PR)
2. #25345 feat(tui): add token activity command [2 of 2]

How to Test

1. Start an app-server client from this branch while authenticated with ChatGPT or AgentIdentity.
2. Call account/usage/read.
3. Confirm the response includes summary and dailyUsageBuckets.
4. Also verify a session without Codex-backend auth receives the existing auth error path.

Targeted tests:

• just test -p codex-backend-client -p codex-app-server-protocol -p codex-app-server
• just write-app-server-schema