Deletes the hand-written internal/tools/mcp client and transports. The
provider talks to an SDK ClientSession over CommandTransport,
StreamableClientTransport or SSEClientTransport, and reconnects when the
server exits. The http transport now keeps a real streamable session
instead of stateless POSTs. Sampling goes through one adapter, since the
SDK marks it deprecated (SEP-2577). message_endpoint is removed: the
SSE transport takes the POST URL from the server's endpoint event.
Model usage was recorded per call and never read, so an org could not see
what it spent. Add the read side:
- model_usage carries identity_id. Task calls take it from the task row,
chat and curator calls from the calling principal carried in the usage
record, so an outbox replay keeps its attribution. Existing task rows are
backfilled from tasks.
- StatsService.UsageTotals sums the caller org's rows by UTC day, source,
provider, model and identity. The org comes from the principal, with no
org parameter, so cross-org reads do not exist.
- GET /api/usage prices tokens at the catalog's list rate, netting cache
hits, and reports calls to unpriced models as tokens only. The published
catalog view now carries each model's cost.
- A new org Usage page behind a read_usage grant: admin and owner only,
with a 7, 30 or 90 day range, totals, the daily series, and model, source
and identity breakdowns. The org shell keeps it on the caller's own org.
A failed RecordUsage was logged and dropped, losing chat, title, sampling
and curator usage that cannot be rebuilt from events. The staterpc Client
now keeps a bounded local outbox: a failed write is queued and retried with
backoff (1s doubling to 1m, 10s per attempt) until the store accepts it, so
records land once the store recovers. The queue is capped at 1024 records;
an overflow is refused so the caller logs the drop. The sampling usage
write is bounded to 5s so a dead store cannot hold a model call.