Report booked cost, entitlement, token coverage, and work as three
series: work per measured token, work per measured+estimated token,
and work per euro. Months below 50% coverage are unfit for token
trends. There is no blended efficiency number.
Split a booked AI-plan fact across workplans or repos using measured
tokens, measured+estimated tokens, session counts, or an even split.
Unmeasured remainder stays a residual. The record is not resource-control
AllocationEvidence and does not create booked spend.
Store mixed measured/estimated/superseded aggregates as session-token
evidence. Map kinds onto the evidence-basis vocabulary, keep unknown
months unknown, and refuse to post the snapshot as booked spend or
resource-control usage.
Add append-only plan entitlements (token, multiplier, or named plan).
Unknown capacity stays unknown and requires a plan label. Plan-month
reporting joins booked euros to entitlement without session logs.
Add ai-plan ingest that stores one current booked fact per
provider-account × plan × period. Subscriptions become recurring
commitments; usage top-ups do not. Session-level rows are rejected.
Record the FIN-WP-0007 split after checking resource-control: they keep
intelligence resource identity and provision-level class I metering;
session tokens stay with State Hub; work-effectiveness is a reporting
join. Missing token counts are now unknown, not zero.
Add create_runway_router() following the hub-core factory pattern,
expose GET /runway/summary and /runway/monthly-burn via create_app,
and add finhub serve for local operation.
Link the DB-first onboarding workstream to FIN-WP-0000-repo-integration.md,
finish agent orientation (rules stubs, remove first-session), add repo
classification, and close all four custodian onboarding tasks.
