* feat(api): add provider quota telemetry, adaptive routing, and status inventory
Adds a read-only OmniRoute status/inventory surface plus supporting
resilience and usage-tracking infrastructure:
- src/lib/quota/providerQuotaTelemetry.ts, providerCapabilities.ts:
provider quota state and capability signals, sourced from configured
metadata rather than invented values; unknown stays unknown.
- src/lib/resilience/adaptiveCircuit.ts, failureClassification.ts:
circuit state with lazy recovery and explicit failure classification.
- src/lib/usage/usageLedger.ts, budgetGuard.ts, modelPricingRegistry.ts:
internal usage tracking and budget allow/warn/deny decisions, kept
separate from upstream-reported quota (never conflated).
- src/lib/routing/adaptiveRouting.ts: excludes exhausted-quota and
open-circuit candidates from routing, penalizes approaching-limit.
- src/lib/omnirouteStatus.ts + src/app/api/omniroute/status,
route/preview: read-only status endpoint; never issues a live
upstream model request (asserted via liveRequestExecuted: false).
- src/lib/db/quotaPools.ts: adds ensurePool() for idempotent pool
management by automation/CLI callers, following the existing
group-demo default-group convention.
- scripts/omniroute-verify.mjs (+ omniroute:verify script): local
verification against the running gateway.
9 new unit tests, all passing. typecheck:core clean relative to base
(release/v3.8.50) -- the 2 pre-existing gateways.ts errors are tracked
separately in #9985 and untouched by this change.
* test(cli): align cli-machine-token assertions with HMAC-SHA256 64-char format
The quota-telemetry feature hardens cliToken to HMAC-SHA256(machineId, SALT)
(64-char hex, pristine machine id). Update the regression test to the new
format and mirror the production derivation in the different-machine-id check.
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
---------
Co-authored-by: adevwithpurpose <adevwithpurpose@users.noreply.github.com>
Co-authored-by: desamours-hub <desamours-hub@users.noreply.github.com>
Co-authored-by: diegosouzapw <8016841+diegosouzapw@users.noreply.github.com>
* fix(cli): derive machine-id token under plain Node and honor salt rotation
`getCliToken()` destructured `machineIdSync` off `await import("node-machine-id")`.
That module is CommonJS, so under plain Node its exports land on `.default` and the
destructured binding is `undefined`. Calling it threw, the bare catch blanked the
token, and every management request went out with no `x-omniroute-cli-token` header
— silently unauthenticated, 401 on every `omniroute combo` / `usage budget` call.
Resolve the binding the same way `src/lib/machineToken.ts` already does, and read
`OMNIROUTE_CLI_SALT` so the rotation documented in docs/security/CLI_TOKEN.md
actually reaches CLI processes (the salt was hardcoded). The catch now logs instead
of failing mute, per the error-handling convention in CONTRIBUTING.md.
The existing test asserted `token === "" || token.length === 32`, so the blanked
token passed. Tightening it in-process is not enough either: the suite runs under
`tsx/esm`, which resolves CJS named exports and hides the bug. The regression test
therefore spawns plain `node` — the loader the CLI actually runs under.
Both new tests fail on the previous code and pass on this one.
* chore(changelog): use the real PR number for the fragment
- 8.2: update-notifier em omniroute.mjs (cache 24h, stderr-only, respeita CI/quiet/json/opt-out)
- 8.12: cliToken.mjs (sha256 de machineId + salt) injetado automaticamente em apiFetch
- 8.12: middleware cliTokenAuth.ts valida token apenas em loopback, timing-safe compare
- 8.12: requireManagementAuth aceita CLI token como bypass local
- env-doc-sync: OMNIROUTE_DISABLE_CLI_TOKEN e OMNIROUTE_NO_UPDATE_NOTIFIER no allowlist
