cartethyia 2.7.4 → 2.7.6
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/.env.example +5 -0
- package/README.md +18 -6
- package/dist/dashboard/assets/BannedPage-B1N603cM.js +1 -0
- package/dist/dashboard/assets/BannedPage-B1N603cM.js.gz +0 -0
- package/dist/dashboard/assets/CliToolDetailPage-9q4_4NoK.js +3 -0
- package/dist/dashboard/assets/CliToolDetailPage-9q4_4NoK.js.gz +0 -0
- package/dist/dashboard/assets/CliToolsPage-DwshGtK-.js +1 -0
- package/dist/dashboard/assets/CliToolsPage-DwshGtK-.js.gz +0 -0
- package/dist/dashboard/assets/CombosPage-D762MIkp.js +1 -0
- package/dist/dashboard/assets/CombosPage-D762MIkp.js.gz +0 -0
- package/dist/dashboard/assets/ConsoleLogPage-KRYjHZsX.js +1 -0
- package/dist/dashboard/assets/ConsoleLogPage-KRYjHZsX.js.gz +0 -0
- package/dist/dashboard/assets/CustomizationPage-DXsxk5uM.js +1 -0
- package/dist/dashboard/assets/CustomizationPage-DXsxk5uM.js.gz +0 -0
- package/dist/dashboard/assets/HealthEventsModal-BEjAjD4F.js +1 -0
- package/dist/dashboard/assets/HealthEventsModal-BEjAjD4F.js.gz +0 -0
- package/dist/dashboard/assets/InjectSystemPromptPanel-d5v3K_Yc.js +899 -0
- package/dist/dashboard/assets/InjectSystemPromptPanel-d5v3K_Yc.js.gz +0 -0
- package/dist/dashboard/assets/LoginPage-Ccv4tjNS.js +1 -0
- package/dist/dashboard/assets/LoginPage-Ccv4tjNS.js.gz +0 -0
- package/dist/dashboard/assets/{ModelPicker-BMS2t8BA.js → ModelPicker-C6WlpU5E.js} +1 -1
- package/dist/dashboard/assets/ModelPicker-C6WlpU5E.js.gz +0 -0
- package/dist/dashboard/assets/OverviewPage-DN9fYNWA.js +1 -0
- package/dist/dashboard/assets/OverviewPage-DN9fYNWA.js.gz +0 -0
- package/dist/dashboard/assets/ProviderDetailPage-BY0ieUgz.js +5 -0
- package/dist/dashboard/assets/ProviderDetailPage-BY0ieUgz.js.gz +0 -0
- package/dist/dashboard/assets/ProviderIcon-jjqXJDA3.js +1 -0
- package/dist/dashboard/assets/ProviderIcon-jjqXJDA3.js.gz +0 -0
- package/dist/dashboard/assets/ProvidersPage-DNsbfea2.js +1 -0
- package/dist/dashboard/assets/ProvidersPage-DNsbfea2.js.gz +0 -0
- package/dist/dashboard/assets/{ProxyPage-lsrx7NYg.js → ProxyPage-DxjvUzSD.js} +1 -1
- package/dist/dashboard/assets/ProxyPage-DxjvUzSD.js.gz +0 -0
- package/dist/dashboard/assets/QuotaPage-OaHScjpX.js +1 -0
- package/dist/dashboard/assets/QuotaPage-OaHScjpX.js.gz +0 -0
- package/dist/dashboard/assets/SettingsPage-BW4AKj2H.js +1 -0
- package/dist/dashboard/assets/SettingsPage-BW4AKj2H.js.gz +0 -0
- package/dist/dashboard/assets/SetupPage-44BZrxFC.js +1 -0
- package/dist/dashboard/assets/SetupPage-44BZrxFC.js.gz +0 -0
- package/dist/dashboard/assets/{StudioPage-Cq5tW7hW.js → StudioPage-DRPavCdU.js} +4 -4
- package/dist/dashboard/assets/StudioPage-DRPavCdU.js.gz +0 -0
- package/dist/dashboard/assets/ToolIcon-CfeLA-Z_.js +1 -0
- package/dist/dashboard/assets/ToolIcon-CfeLA-Z_.js.gz +0 -0
- package/dist/dashboard/assets/UsagePage-B7M7LvMM.js +56 -0
- package/dist/dashboard/assets/UsagePage-B7M7LvMM.js.gz +0 -0
- package/dist/dashboard/assets/api-keys-KvX83mRe.js +1 -0
- package/dist/dashboard/assets/api-keys-KvX83mRe.js.gz +0 -0
- package/dist/dashboard/assets/app-DRnQLhsc.js +9 -0
- package/dist/dashboard/assets/app-DRnQLhsc.js.gz +0 -0
- package/dist/dashboard/assets/app-mweRcsKp.css +2 -0
- package/dist/dashboard/assets/app-mweRcsKp.css.gz +0 -0
- package/dist/dashboard/assets/assert-CidQVoNo.js +1 -0
- package/dist/dashboard/assets/assert-CidQVoNo.js.gz +0 -0
- package/dist/dashboard/assets/{badge-D9CS6dZk.js → badge-BaX2GTrn.js} +1 -1
- package/dist/dashboard/assets/badge-BaX2GTrn.js.gz +0 -0
- package/dist/dashboard/assets/{banner-DympBNq4.js → banner-DFH4Bkte.js} +1 -1
- package/dist/dashboard/assets/banner-DFH4Bkte.js.gz +0 -0
- package/dist/dashboard/assets/brand/login_register.jpg +0 -0
- package/dist/dashboard/assets/brand/login_register_initial.jpg +0 -0
- package/dist/dashboard/assets/brand/og_bansos.jpg +0 -0
- package/dist/dashboard/assets/brand/og_image.jpg +0 -0
- package/dist/dashboard/assets/{card-ibRWU_uT.js → card-CxDw0lK6.js} +1 -1
- package/dist/dashboard/assets/card-CxDw0lK6.js.gz +0 -0
- package/dist/dashboard/assets/chevron-left-Bf6CaTSJ.js +1 -0
- package/dist/dashboard/assets/chevron-left-Bf6CaTSJ.js.gz +0 -0
- package/dist/dashboard/assets/circle-question-mark-Cz4un8sH.js +1 -0
- package/dist/dashboard/assets/circle-question-mark-Cz4un8sH.js.gz +0 -0
- package/dist/dashboard/assets/cli-tools-Bwv5-zGZ.js +1 -0
- package/dist/dashboard/assets/cli-tools-Bwv5-zGZ.js.gz +0 -0
- package/dist/dashboard/assets/{clipboard-button-DJm-9pz7.js → clipboard-button-DfS1ntiq.js} +1 -1
- package/dist/dashboard/assets/clipboard-button-DfS1ntiq.js.gz +0 -0
- package/dist/dashboard/assets/dialog-BK-NRr5u.js +1 -0
- package/dist/dashboard/assets/dialog-BK-NRr5u.js.gz +0 -0
- package/dist/dashboard/assets/{drawer-CqGYeOrd.js → drawer-QqPul6sC.js} +1 -1
- package/dist/dashboard/assets/drawer-QqPul6sC.js.gz +0 -0
- package/dist/dashboard/assets/entry-BMlYJVFj.js +1 -0
- package/dist/dashboard/assets/entry-BMlYJVFj.js.gz +0 -0
- package/dist/dashboard/assets/entry-D6WvojeD.js +4 -0
- package/dist/dashboard/assets/entry-D6WvojeD.js.gz +0 -0
- package/dist/dashboard/assets/entry-Dmx7gBiN.js +1 -0
- package/dist/dashboard/assets/entry-Dmx7gBiN.js.gz +0 -0
- package/dist/dashboard/assets/{formatters-D-MTJxMr.js → formatters-u7JPB5BX.js} +1 -1
- package/dist/dashboard/assets/formatters-u7JPB5BX.js.gz +0 -0
- package/dist/dashboard/assets/house-h9AZlJuT.js +1 -0
- package/dist/dashboard/assets/house-h9AZlJuT.js.gz +0 -0
- package/dist/dashboard/assets/{index-DBNPOfhV.js → index-Ko4dLRnV.js} +2 -2
- package/dist/dashboard/assets/index-Ko4dLRnV.js.gz +0 -0
- package/dist/dashboard/assets/{inline-BT0zntEr.js → inline-CHJQwxLa.js} +1 -1
- package/dist/dashboard/assets/inline-CHJQwxLa.js.gz +0 -0
- package/dist/dashboard/assets/{input-Da2ZViWg.js → input-BlmlizVV.js} +1 -1
- package/dist/dashboard/assets/input-BlmlizVV.js.gz +0 -0
- package/dist/dashboard/assets/{layout-BXNeLw6M.js → layout-Bz93I-yE.js} +1 -1
- package/dist/dashboard/assets/layout-Bz93I-yE.js.gz +0 -0
- package/dist/dashboard/assets/{list-row-BycOrrIV.js → list-row-Ml16KCfu.js} +1 -1
- package/dist/dashboard/assets/list-row-Ml16KCfu.js.gz +0 -0
- package/dist/dashboard/assets/live-7_QpnYGu.js +1 -0
- package/dist/dashboard/assets/live-7_QpnYGu.js.gz +0 -0
- package/dist/dashboard/assets/{mutation-Rr0TIvcj.js → mutation-GU7XdEjz.js} +1 -1
- package/dist/dashboard/assets/mutation-GU7XdEjz.js.gz +0 -0
- package/dist/dashboard/assets/network-CuMDaTzT.js +1 -0
- package/dist/dashboard/assets/network-CuMDaTzT.js.gz +0 -0
- package/dist/dashboard/assets/numbers-CoqpaDMx.js +2 -0
- package/dist/dashboard/assets/numbers-CoqpaDMx.js.gz +0 -0
- package/dist/dashboard/assets/pagination-B5aOoweZ.js +1 -0
- package/dist/dashboard/assets/pagination-B5aOoweZ.js.gz +0 -0
- package/dist/dashboard/assets/providers/zai.webp +0 -0
- package/dist/dashboard/assets/providers/zcode.webp +0 -0
- package/dist/dashboard/assets/{query-keys-BloKtaXJ.js → query-keys-YSoQOUh6.js} +1 -1
- package/dist/dashboard/assets/{query-keys-BloKtaXJ.js.gz → query-keys-YSoQOUh6.js.gz} +0 -0
- package/dist/dashboard/assets/quota-DJ1QlAOq.js +1 -0
- package/dist/dashboard/assets/quota-DJ1QlAOq.js.gz +0 -0
- package/dist/dashboard/assets/routing-fnCsxgNO.js +1 -0
- package/dist/dashboard/assets/routing-fnCsxgNO.js.gz +0 -0
- package/dist/dashboard/assets/{select-BIYIBkMU.js → select-BHqpd-bg.js} +2 -2
- package/dist/dashboard/assets/select-BHqpd-bg.js.gz +0 -0
- package/dist/dashboard/assets/settings-BxslsErn.js +1 -0
- package/dist/dashboard/assets/settings-BxslsErn.js.gz +0 -0
- package/dist/dashboard/assets/shield-BMQNxORh.js +1 -0
- package/dist/dashboard/assets/shield-BMQNxORh.js.gz +0 -0
- package/dist/dashboard/assets/{stack-DfgnFDWY.js → stack-j7cXjgYX.js} +1 -1
- package/dist/dashboard/assets/stack-j7cXjgYX.js.gz +0 -0
- package/dist/dashboard/assets/{state-CkfdYWel.js → state-DHmEnLa2.js} +1 -1
- package/dist/dashboard/assets/state-DHmEnLa2.js.gz +0 -0
- package/dist/dashboard/assets/{switch-SYm237HG.js → switch-DmZcyTIC.js} +1 -1
- package/dist/dashboard/assets/switch-DmZcyTIC.js.gz +0 -0
- package/dist/dashboard/assets/{useMutation-BQqCRNIi.js → useMutation-CbCP4w1o.js} +1 -1
- package/dist/dashboard/assets/useMutation-CbCP4w1o.js.gz +0 -0
- package/dist/dashboard/console-backgrounds.jpg +0 -0
- package/dist/dashboard/console-backgrounds_initial.jpg +0 -0
- package/dist/dashboard/index.html +2 -2
- package/dist/dashboard/index.html.gz +0 -0
- package/dist/dashboard/precompressed-assets.json +60 -59
- package/migrations/full-postgres/0036_system_prompt_presets_refs.sql +10 -0
- package/migrations/full-postgres/0037_share_concurrency_scope.sql +3 -0
- package/migrations/full-postgres/0038_add_accounts_quota_snapshot.sql +7 -0
- package/migrations/full-postgres/0039_add_telemetry_provider_model_indexes.sql +13 -0
- package/migrations/full-postgres/0040_add_models_dialect.sql +3 -0
- package/migrations/full-postgres/0041_add_daily_checkin_ledger_account_fk.sql +7 -0
- package/migrations/full-postgres/0042_add_telemetry_degenerate_stop.sql +14 -0
- package/migrations/lite-sqlite3/0036_system_prompt_presets_refs.sql +14 -0
- package/migrations/lite-sqlite3/0037_share_concurrency_scope.sql +3 -0
- package/migrations/lite-sqlite3/0038_add_accounts_quota_snapshot.sql +4 -0
- package/migrations/lite-sqlite3/0039_add_telemetry_provider_model_indexes.sql +6 -0
- package/migrations/lite-sqlite3/0040_add_models_dialect.sql +3 -0
- package/migrations/lite-sqlite3/0041_add_daily_checkin_ledger_account_fk.sql +26 -0
- package/migrations/lite-sqlite3/0042_add_telemetry_degenerate_stop.sql +12 -0
- package/package.json +4 -6
- package/scripts/cli/update.ts +15 -41
- package/src/boot/app.ts +58 -2
- package/src/boot/dependencies.ts +794 -732
- package/src/boot/lifecycle.ts +11 -12
- package/src/config/cooldowns.ts +34 -55
- package/src/config/network.ts +1 -1
- package/src/config/package-paths.ts +1 -1
- package/src/config/providers.ts +19 -14
- package/src/config/resources.ts +0 -5
- package/src/config/runtime.ts +2 -5
- package/src/config/spec.ts +33 -6
- package/src/config/usage.ts +27 -0
- package/src/console/api-keys/ban-constants.ts +38 -0
- package/src/console/api-keys/contracts.ts +71 -62
- package/src/console/api-keys/routes.ts +56 -24
- package/src/console/auth/access.ts +4 -4
- package/src/console/auth/account-reset.ts +1 -1
- package/src/console/auth/audit-recorder.ts +1 -1
- package/src/console/auth/lockout.ts +7 -3
- package/src/console/auth/row-coercion.ts +57 -12
- package/src/console/auth/session-resolver.ts +1 -1
- package/src/console/auth/session-service.ts +8 -2
- package/src/console/auth/session.ts +37 -31
- package/src/console/auth/setup-service.ts +4 -3
- package/src/console/auth/user-store.ts +1 -1
- package/src/console/backup/9router-map.ts +0 -3
- package/src/console/backup/contracts.ts +9 -9
- package/src/console/backup/export-endpoint.ts +5 -6
- package/src/console/backup/nine-router.ts +1 -7
- package/src/console/backup/routes.ts +29 -33
- package/src/console/backup/service.ts +468 -468
- package/src/console/backup/store.ts +47 -29
- package/src/console/backup/validate.ts +11 -1
- package/src/console/cli-tools/contracts.ts +43 -299
- package/src/console/cli-tools/fs-ops.ts +29 -0
- package/src/console/cli-tools/injectors/claude.ts +3 -3
- package/src/console/cli-tools/injectors/codex.ts +181 -181
- package/src/console/cli-tools/injectors/driver.ts +150 -247
- package/src/console/cli-tools/injectors/kilo.ts +146 -21
- package/src/console/cli-tools/injectors/opencode.ts +176 -153
- package/src/console/cli-tools/routes.ts +7 -1
- package/src/console/cli-tools/service.ts +8 -7
- package/src/console/cli-tools/store.ts +2 -2
- package/src/console/console-router.ts +32 -0
- package/src/console/dashboard-assets.ts +9 -9
- package/src/console/domain-registration.ts +16 -2
- package/src/console/inferhub/client.ts +0 -7
- package/src/console/inferhub/key-store.ts +1 -1
- package/src/console/inferhub/normalize.ts +2 -2
- package/src/console/inferhub/routes.ts +1 -1
- package/src/console/inferhub/types.ts +1 -1
- package/src/console/observability/contracts.ts +7 -1
- package/src/console/observability/live.ts +31 -18
- package/src/console/observability/logs.ts +1 -1
- package/src/console/observability/sse.ts +62 -8
- package/src/console/observability/store.ts +23 -4
- package/src/console/providers/catalog/catalog-projections.ts +1 -1
- package/src/console/providers/catalog/contracts.ts +39 -0
- package/src/console/providers/catalog/model-operations.ts +1 -1
- package/src/console/providers/catalog/provider-operations.ts +95 -17
- package/src/console/providers/catalog/public-model-store.ts +0 -0
- package/src/console/providers/catalog/routes.ts +27 -3
- package/src/console/providers/catalog/store.ts +147 -8
- package/src/console/providers/detail/routes.ts +1 -1
- package/src/console/providers/oauth/callback-listener.ts +100 -18
- package/src/console/providers/oauth/routes.ts +92 -13
- package/src/console/quota/account-quota-global.ts +41 -3
- package/src/console/quota/account-quota-shared.ts +10 -3
- package/src/console/quota/account-quota-tenant.ts +259 -66
- package/src/console/quota/daily-growth-job.ts +308 -0
- package/src/console/quota/native-identity.ts +2 -2
- package/src/console/quota/quota-view.ts +1 -1
- package/src/console/remote-support/routes.ts +1 -1
- package/src/console/remote-support/service.ts +2 -2
- package/src/console/routing/model/contracts.ts +1 -1
- package/src/console/routing/pools/contracts.ts +2 -2
- package/src/console/routing/pools/import-parse.ts +3 -3
- package/src/console/routing/pools/relay-deploy.ts +3 -3
- package/src/console/settings/contracts.ts +29 -2
- package/src/console/settings/store.ts +20 -0
- package/src/console/share/share-router.ts +100 -32
- package/src/console/share/share-stats.ts +11 -13
- package/src/console/share/share-usage.ts +1 -1
- package/src/console/shared/errors.ts +2 -2
- package/src/console/shared/object.ts +18 -0
- package/src/console/shared/read-through-cache.ts +4 -4
- package/src/console/shared/transaction.ts +14 -0
- package/src/console/system-prompt/contracts.ts +89 -12
- package/src/console/system-prompt/routes.ts +98 -45
- package/src/console/updates/apply-update.ts +244 -0
- package/src/console/updates/routes.ts +252 -97
- package/src/main.ts +11 -5
- package/src/network/http2-bidi.ts +1 -1
- package/src/network/ip-utils.ts +298 -0
- package/src/network/outbound-fetch.ts +1 -1
- package/src/network/pool/agent.ts +7 -6
- package/src/network/pool/byte-accounting.ts +1 -1
- package/src/network/pool/resolver.ts +16 -12
- package/src/network/pool/selector.ts +150 -136
- package/src/network/pool-health-machine.ts +1 -1
- package/src/network/rate-limit-verdict.ts +1 -1
- package/src/network/retry.ts +1 -1
- package/src/network/ssrf.ts +20 -71
- package/src/observability/event-log.ts +9 -2
- package/src/observability/event-query.ts +61 -291
- package/src/observability/log-sanitize.ts +1 -1
- package/src/observability/metrics.ts +4 -4
- package/src/observability/payload-capture.ts +6 -5
- package/src/observability/payload-store.ts +1 -1
- package/src/observability/runtime-metrics.ts +1 -1
- package/src/observability/telemetry-buffer.ts +19 -3
- package/src/observability/telemetry-reconciliation.ts +3 -3
- package/src/observability/token-speed.ts +1 -1
- package/src/persistence/connection-url.ts +1 -1
- package/src/persistence/db-handle.ts +1 -1
- package/src/persistence/db-pg.ts +2 -2
- package/src/persistence/db-sqlite.ts +3 -3
- package/src/persistence/dialect.ts +5 -1
- package/src/persistence/readiness.ts +1 -1
- package/src/persistence/redis.ts +1 -1
- package/src/persistence/schema.ts +121 -40
- package/src/persistence/shared-cache.ts +2 -2
- package/src/persistence/sort-order.ts +1 -1
- package/src/persistence/stores/api-key-store.ts +29 -2
- package/src/persistence/stores/share-store.ts +20 -8
- package/src/persistence/system-prompt-store.ts +63 -18
- package/src/persistence/tenant-preferences.ts +1 -1
- package/src/persistence/ttl-cache.ts +3 -3
- package/src/persistence/usage-flush.ts +7 -2
- package/src/protocol/canonical-model.ts +1 -1
- package/src/protocol/codecs.ts +6 -10
- package/src/protocol/dsml.ts +8 -125
- package/src/protocol/image-clamp.ts +2 -2
- package/src/protocol/reasoning-text.ts +1 -1
- package/src/protocol/reasoning.ts +7 -52
- package/src/protocol/registry.ts +50 -15
- package/src/protocol/request/chat.ts +46 -19
- package/src/protocol/request/codex.ts +1 -1
- package/src/protocol/request/messages.ts +46 -14
- package/src/protocol/request/responses.ts +19 -10
- package/src/protocol/response/chat.ts +25 -10
- package/src/protocol/response/codex.ts +1 -1
- package/src/protocol/response/messages.ts +1 -1
- package/src/protocol/response/omp-tool-call-markup.ts +1 -1
- package/src/protocol/response/responses.ts +1 -1
- package/src/protocol/sse-decoder.ts +50 -1
- package/src/protocol/stop-reason.ts +2 -2
- package/src/protocol/think-split.ts +347 -0
- package/src/protocol/translation/attachment-integrity.ts +3 -3
- package/src/protocol/translation/cache-controls.ts +8 -20
- package/src/protocol/translation/capabilities.ts +6 -6
- package/src/protocol/translation/dialect/deepseek.ts +10 -0
- package/src/protocol/translation/dialect/inband.ts +39 -0
- package/src/protocol/translation/dialect/model-thinking.ts +89 -0
- package/src/protocol/translation/dialect/qwen.ts +114 -0
- package/src/protocol/translation/dialect/resolver.ts +83 -0
- package/src/protocol/translation/dialect/rules.ts +93 -0
- package/src/protocol/translation/dialect/thinking-ladder.ts +245 -0
- package/src/protocol/translation/dialect/types.ts +58 -0
- package/src/protocol/translation/quirks.ts +18 -4
- package/src/protocol/translation/thinking.ts +12 -252
- package/src/protocol/translation/tool-id.ts +3 -3
- package/src/protocol/translation/tool-repair.ts +1 -1
- package/src/protocol/upstream/post-json.ts +6 -1
- package/src/providers/adapters/compatible-adapter.ts +71 -4
- package/src/providers/authentication/jwt-validator.ts +0 -5
- package/src/providers/authentication/oauth-client.ts +254 -253
- package/src/providers/authentication/oauth-flow-store.ts +42 -7
- package/src/providers/bundled-discovery-providers.ts +6 -30
- package/src/providers/configured-openai-providers.ts +82 -97
- package/src/providers/default-registry.ts +35 -37
- package/src/providers/discovery/catalog/catalog-format.ts +13 -2
- package/src/providers/discovery/discovery-types.ts +0 -1
- package/src/providers/discovery/openai-model-discovery.ts +2 -2
- package/src/providers/discovery/probe-phases.ts +48 -13
- package/src/providers/discovery/probing-service.ts +6 -8
- package/src/providers/integrations/agentrouter/agentrouter.ts +12 -2
- package/src/providers/integrations/antigravity/antigravity-oauth.ts +0 -1
- package/src/providers/integrations/antigravity/antigravity-protocol.ts +88 -28
- package/src/providers/integrations/antigravity/antigravity-quota.ts +18 -6
- package/src/providers/integrations/antigravity/antigravity.ts +22 -13
- package/src/providers/integrations/buddy/buddy-catalog-shared.ts +1 -1
- package/src/providers/integrations/buddy/buddy-chat-shared.ts +20 -4
- package/src/providers/integrations/buddy/buddy-cli-headers.ts +19 -52
- package/src/providers/integrations/buddy/buddy-identity.ts +86 -0
- package/src/providers/integrations/buddy/buddy-oauth-shared.ts +1 -1
- package/src/providers/integrations/buddy/codebuddy/cn-provider.ts +1 -2
- package/src/providers/integrations/buddy/codebuddy/provider.ts +4 -15
- package/src/providers/integrations/buddy/codebuddy/shared.ts +30 -9
- package/src/providers/integrations/buddy/workbuddy/provider.ts +5 -9
- package/src/providers/integrations/buddy/workbuddy/shared.ts +21 -3
- package/src/providers/integrations/claude/claude-cch.ts +148 -148
- package/src/providers/integrations/claude/fallback-credit.ts +3 -7
- package/src/providers/integrations/cline/cline.ts +1 -1
- package/src/providers/integrations/codex/codex-device-code.ts +101 -101
- package/src/providers/integrations/codex/codex-oauth.ts +0 -1
- package/src/providers/integrations/commandcode/commandcode.ts +75 -115
- package/src/providers/integrations/commandcode/quota.ts +131 -51
- package/src/providers/integrations/cursor/cursor-http2.ts +0 -3
- package/src/providers/integrations/cursor/cursor-quota.ts +2 -6
- package/src/providers/integrations/cursor/cursor-request.ts +0 -87
- package/src/providers/integrations/cursor/cursor-shared.ts +0 -4
- package/src/providers/integrations/cursor/cursor.ts +35 -3
- package/src/providers/integrations/cursor/protocol/framing.ts +0 -56
- package/src/providers/integrations/cursor/protocol/messages.ts +0 -96
- package/src/providers/integrations/factory/factory.ts +9 -9
- package/src/providers/integrations/github/github-discovery.ts +70 -12
- package/src/providers/integrations/github/github-oauth.ts +184 -14
- package/src/providers/integrations/github/github.ts +134 -26
- package/src/providers/integrations/grok/grok-oauth.ts +261 -261
- package/src/providers/integrations/grok/grok-quota-frame.ts +1 -1
- package/src/providers/integrations/grok/grok-quota.ts +16 -7
- package/src/providers/integrations/grok/grok-ratelimit-store.ts +0 -5
- package/src/providers/integrations/grok/grok-ratelimit.ts +3 -0
- package/src/providers/integrations/grok/grok.ts +86 -36
- package/src/providers/integrations/kilo/kilo-oauth.ts +1 -1
- package/src/providers/integrations/kiro/kiro-cache.ts +31 -0
- package/src/providers/integrations/kiro/kiro-catalog.ts +31 -5
- package/src/providers/integrations/kiro/kiro-discovery.ts +47 -43
- package/src/providers/integrations/kiro/kiro-machine-id.ts +6 -3
- package/src/providers/integrations/kiro/kiro-oauth.ts +523 -255
- package/src/providers/integrations/kiro/kiro-quota.ts +35 -47
- package/src/providers/integrations/kiro/kiro-request.ts +8 -11
- package/src/providers/integrations/kiro/kiro-stream.ts +13 -0
- package/src/providers/integrations/kiro/kiro.ts +80 -68
- package/src/providers/integrations/muse/muse-oauth.ts +0 -1
- package/src/providers/integrations/muse/muse-quota.ts +0 -4
- package/src/providers/integrations/opencode/fingerprint.ts +7 -5
- package/src/providers/integrations/opencode/provider.ts +17 -24
- package/src/providers/integrations/qoder/qoder-catalog.ts +94 -11
- package/src/providers/integrations/qoder/qoder-oauth.ts +165 -25
- package/src/providers/integrations/qoder/qoder-quota.ts +48 -6
- package/src/providers/integrations/qoder/qoder-realm.ts +10 -31
- package/src/providers/integrations/qoder/qoder-sse.ts +159 -0
- package/src/providers/integrations/qoder/qoder-umid.ts +21 -25
- package/src/providers/integrations/qoder/qoder-virtual.ts +12 -0
- package/src/providers/integrations/qoder/qoder.ts +422 -71
- package/src/providers/integrations/shared/connect.ts +0 -54
- package/src/providers/integrations/shared/hash.ts +15 -0
- package/src/providers/integrations/shared/session-cache.ts +136 -0
- package/src/providers/integrations/trae/trae-cache.ts +7 -87
- package/src/providers/integrations/trae/trae-catalog.ts +1 -6
- package/src/providers/integrations/trae/trae-protocol.ts +17 -8
- package/src/providers/integrations/trae/trae-realm.ts +0 -7
- package/src/providers/integrations/trae/trae.ts +51 -8
- package/src/providers/integrations/xai/xai-oauth.ts +5 -5
- package/src/providers/integrations/zcode-start-plan/gifts.ts +265 -0
- package/src/providers/integrations/zcode-start-plan/oauth.ts +154 -0
- package/src/providers/integrations/zcode-start-plan/provider.ts +131 -0
- package/src/providers/integrations/zcode-start-plan/quota.ts +152 -0
- package/src/providers/integrations/zcode-start-plan/zcode_prompt.json +10 -0
- package/src/providers/operations/account-error-classification.ts +198 -133
- package/src/providers/operations/account-health-service.ts +31 -39
- package/src/providers/operations/account-reset-service.ts +58 -1
- package/src/providers/operations/byok-wire-profile.ts +19 -15
- package/src/providers/operations/cli-platform.ts +31 -5
- package/src/providers/operations/client-version-resolver.ts +4 -6
- package/src/providers/operations/client-versions.ts +10 -7
- package/src/providers/operations/provider-catalog-seeder.ts +2 -0
- package/src/providers/operations/provider-catalog-service.ts +18 -2
- package/src/providers/operations/provider-credential-service.ts +19 -2
- package/src/providers/provider-contracts.ts +18 -7
- package/src/providers/provider-metadata.ts +17 -2
- package/src/providers/provider-model-definition.ts +22 -1
- package/src/providers/provider-probe.ts +4 -2
- package/src/providers/provider-usage.ts +50 -19
- package/src/providers/quota/cache.ts +29 -0
- package/src/providers/quota/quota-contracts.ts +4 -0
- package/src/providers/quota/refresh.ts +27 -4
- package/src/providers/quota/snapshot.ts +94 -0
- package/src/providers/search/search-provider.ts +1 -3
- package/src/providers/search/search-providers.ts +2 -1
- package/src/security/abuse.ts +4 -4
- package/src/security/admission/service.ts +13 -6
- package/src/security/admission/ttl.ts +61 -61
- package/src/security/api-key-auth.ts +17 -3
- package/src/security/client-router-fingerprint.ts +4 -4
- package/src/security/credential-codec.ts +3 -3
- package/src/security/csrf.ts +2 -2
- package/src/security/ip-boundary.ts +16 -189
- package/src/security/model-access-rule.ts +1 -1
- package/src/security/outbound-headers.ts +18 -0
- package/src/shared/invoke-guard.ts +47 -0
- package/src/transport/contracts.ts +13 -4
- package/src/transport/dispatch/attempt-finalize.ts +70 -22
- package/src/transport/dispatch/attempt-loop.ts +22 -0
- package/src/transport/dispatch/capture-policy.ts +34 -37
- package/src/transport/dispatch/empty-stream.ts +23 -20
- package/src/transport/dispatch/generation.ts +25 -45
- package/src/transport/dispatch/native-shared.ts +127 -0
- package/src/transport/dispatch/proxy-request.ts +3 -1
- package/src/transport/dispatch/responses-compact.ts +25 -46
- package/src/transport/dispatch/streaming/encoder.ts +18 -72
- package/src/transport/dispatch/streaming/finalize.ts +2 -2
- package/src/transport/dispatch/streaming/stream.ts +7 -3
- package/src/transport/dispatch/streaming/thinking-loop.ts +1 -1
- package/src/transport/dispatch/systemone.ts +20 -48
- package/src/transport/dispatch/token-count.ts +41 -38
- package/src/transport/dispatch/websearch.ts +19 -45
- package/src/transport/drain-endpoint.ts +1 -1
- package/src/transport/failure-policy.ts +21 -1
- package/src/transport/guards/auth-guard.ts +0 -6
- package/src/transport/guards/body-policy.ts +3 -3
- package/src/transport/guards/error-lifecycle.ts +4 -0
- package/src/transport/guards/headroom-guard.ts +1 -1
- package/src/transport/guards/pipeline.ts +1 -1
- package/src/transport/guards/request-context.ts +12 -17
- package/src/transport/request/preparer.ts +286 -7
- package/src/transport/request/rtk/compress-request.ts +3 -3
- package/src/transport/request/rtk/rtk-constants.ts +2 -2
- package/src/transport/request/rtk/rtk-filters.ts +8 -8
- package/src/transport/request/state.ts +7 -13
- package/src/transport/request/system-prompt-injection.ts +74 -11
- package/src/transport/request/tenant-preferences.ts +22 -0
- package/src/transport/routing/candidate-index.ts +2 -2
- package/src/transport/routing/round-robin.ts +1 -1
- package/src/transport/routing/route-catalog.ts +17 -18
- package/src/transport/routing/route-model.ts +126 -33
- package/src/transport/routing/router.ts +159 -37
- package/src/transport/shutdown-notice.ts +2 -2
- package/src/transport/surface/adapters.ts +3 -3
- package/src/transport/surface/chat/encode.ts +7 -7
- package/src/transport/surface/chat/parse.ts +7 -7
- package/src/transport/surface/completion.ts +2 -2
- package/src/transport/surface/content-parts.ts +1 -1
- package/src/transport/surface/dialects.ts +2 -2
- package/src/transport/surface/messages/parse.ts +3 -3
- package/src/transport/surface/reasoning-summary.ts +9 -14
- package/src/transport/surface/responses/adapter.ts +1 -1
- package/src/transport/surface/responses/errors.ts +0 -9
- package/src/transport/vision-handoff-cache.ts +58 -0
- package/src/transport/vision-handoff.ts +431 -0
- package/src/workers/account-identity-sweep.ts +2 -2
- package/src/workers/buddy-growth-tasks.ts +392 -346
- package/src/workers/daily-checkin.ts +102 -49
- package/src/workers/maintenance-lease.ts +2 -35
- package/src/workers/oauth-refresh-sweep.ts +8 -1
- package/src/workers/quota-refresh-sweep.ts +120 -82
- package/src/workers/scheduler.ts +67 -4
- package/src/workers/sweep-runner.ts +24 -5
- package/src/workers/zcode-gift-claim.ts +108 -0
- package/dist/dashboard/assets/AccountCooldown-Bjo92Qat.js +0 -1
- package/dist/dashboard/assets/AccountCooldown-Bjo92Qat.js.gz +0 -0
- package/dist/dashboard/assets/BannedPage-ChG5svFb.js +0 -1
- package/dist/dashboard/assets/BannedPage-ChG5svFb.js.gz +0 -0
- package/dist/dashboard/assets/CliToolDetailPage-BIcWQun7.js +0 -1
- package/dist/dashboard/assets/CliToolDetailPage-BIcWQun7.js.gz +0 -0
- package/dist/dashboard/assets/CliToolsPage-Bf2VelDe.js +0 -1
- package/dist/dashboard/assets/CliToolsPage-Bf2VelDe.js.gz +0 -0
- package/dist/dashboard/assets/CombosPage-DCgm5g6g.js +0 -1
- package/dist/dashboard/assets/CombosPage-DCgm5g6g.js.gz +0 -0
- package/dist/dashboard/assets/ConsoleLogPage-a8mD0QF1.js +0 -1
- package/dist/dashboard/assets/ConsoleLogPage-a8mD0QF1.js.gz +0 -0
- package/dist/dashboard/assets/CustomizationPage-ggeMSjte.js +0 -1
- package/dist/dashboard/assets/CustomizationPage-ggeMSjte.js.gz +0 -0
- package/dist/dashboard/assets/HealthEventsModal-DzSWzE_j.js +0 -1
- package/dist/dashboard/assets/HealthEventsModal-DzSWzE_j.js.gz +0 -0
- package/dist/dashboard/assets/LoginPage-U6AKaJcF.js +0 -1
- package/dist/dashboard/assets/LoginPage-U6AKaJcF.js.gz +0 -0
- package/dist/dashboard/assets/ModelPicker-BMS2t8BA.js.gz +0 -0
- package/dist/dashboard/assets/OverviewPage-D2D4IiHl.js +0 -1
- package/dist/dashboard/assets/OverviewPage-D2D4IiHl.js.gz +0 -0
- package/dist/dashboard/assets/ProviderDetailPage-CrNf7j_X.js +0 -5
- package/dist/dashboard/assets/ProviderDetailPage-CrNf7j_X.js.gz +0 -0
- package/dist/dashboard/assets/ProviderIcon-FB6yS4v1.js +0 -1
- package/dist/dashboard/assets/ProviderIcon-FB6yS4v1.js.gz +0 -0
- package/dist/dashboard/assets/ProvidersPage-BSDbquc1.js +0 -1
- package/dist/dashboard/assets/ProvidersPage-BSDbquc1.js.gz +0 -0
- package/dist/dashboard/assets/ProxyPage-lsrx7NYg.js.gz +0 -0
- package/dist/dashboard/assets/QuotaPage-L0skzML8.js +0 -1
- package/dist/dashboard/assets/QuotaPage-L0skzML8.js.gz +0 -0
- package/dist/dashboard/assets/SettingsPage-CmAbjRHW.js +0 -1
- package/dist/dashboard/assets/SettingsPage-CmAbjRHW.js.gz +0 -0
- package/dist/dashboard/assets/SetupPage-f0DE1h0-.js +0 -1
- package/dist/dashboard/assets/SetupPage-f0DE1h0-.js.gz +0 -0
- package/dist/dashboard/assets/SortableList-ClzpwyET.js +0 -1
- package/dist/dashboard/assets/SortableList-ClzpwyET.js.gz +0 -0
- package/dist/dashboard/assets/StudioPage-Cq5tW7hW.js.gz +0 -0
- package/dist/dashboard/assets/ToolIcon-K1ckcDKg.js +0 -1
- package/dist/dashboard/assets/ToolIcon-K1ckcDKg.js.gz +0 -0
- package/dist/dashboard/assets/UsagePage-Dfc9rWQn.js +0 -56
- package/dist/dashboard/assets/UsagePage-Dfc9rWQn.js.gz +0 -0
- package/dist/dashboard/assets/api-keys-C38iXoO4.js +0 -1
- package/dist/dashboard/assets/api-keys-C38iXoO4.js.gz +0 -0
- package/dist/dashboard/assets/app-BayTTy01.js +0 -9
- package/dist/dashboard/assets/app-BayTTy01.js.gz +0 -0
- package/dist/dashboard/assets/app-a159k0ub.css +0 -2
- package/dist/dashboard/assets/app-a159k0ub.css.gz +0 -0
- package/dist/dashboard/assets/assert-DbRA8EG2.js +0 -15
- package/dist/dashboard/assets/assert-DbRA8EG2.js.gz +0 -0
- package/dist/dashboard/assets/badge-D9CS6dZk.js.gz +0 -0
- package/dist/dashboard/assets/banner-DympBNq4.js.gz +0 -0
- package/dist/dashboard/assets/card-ibRWU_uT.js.gz +0 -0
- package/dist/dashboard/assets/cli-tools-63qBGaRw.js +0 -1
- package/dist/dashboard/assets/cli-tools-63qBGaRw.js.gz +0 -0
- package/dist/dashboard/assets/clipboard-button-DJm-9pz7.js.gz +0 -0
- package/dist/dashboard/assets/dialog-8fiE6tec.js +0 -1
- package/dist/dashboard/assets/dialog-8fiE6tec.js.gz +0 -0
- package/dist/dashboard/assets/drawer-CqGYeOrd.js.gz +0 -0
- package/dist/dashboard/assets/entry-BSAsYsmV.js +0 -1
- package/dist/dashboard/assets/entry-BSAsYsmV.js.gz +0 -0
- package/dist/dashboard/assets/entry-CiIQULZD.js +0 -4
- package/dist/dashboard/assets/entry-CiIQULZD.js.gz +0 -0
- package/dist/dashboard/assets/entry-Daz0lJ-W.js +0 -1
- package/dist/dashboard/assets/entry-Daz0lJ-W.js.gz +0 -0
- package/dist/dashboard/assets/formatters-D-MTJxMr.js.gz +0 -0
- package/dist/dashboard/assets/github-badge-CPK47xQx.js +0 -1
- package/dist/dashboard/assets/github-badge-CPK47xQx.js.gz +0 -0
- package/dist/dashboard/assets/index-DBNPOfhV.js.gz +0 -0
- package/dist/dashboard/assets/inline-BT0zntEr.js.gz +0 -0
- package/dist/dashboard/assets/input-Da2ZViWg.js.gz +0 -0
- package/dist/dashboard/assets/layout-BXNeLw6M.js.gz +0 -0
- package/dist/dashboard/assets/list-row-BycOrrIV.js.gz +0 -0
- package/dist/dashboard/assets/live-DCQNZpF7.js +0 -1
- package/dist/dashboard/assets/live-DCQNZpF7.js.gz +0 -0
- package/dist/dashboard/assets/mutation-Rr0TIvcj.js.gz +0 -0
- package/dist/dashboard/assets/network-9Yj2ZyLo.js +0 -1
- package/dist/dashboard/assets/network-9Yj2ZyLo.js.gz +0 -0
- package/dist/dashboard/assets/numbers-CEagj6ZZ.js +0 -2
- package/dist/dashboard/assets/numbers-CEagj6ZZ.js.gz +0 -0
- package/dist/dashboard/assets/providers/zai-coding-plan.svg +0 -6
- package/dist/dashboard/assets/providers/zai-coding-plan.svg.gz +0 -0
- package/dist/dashboard/assets/quota-Caw7wqci.js +0 -1
- package/dist/dashboard/assets/quota-Caw7wqci.js.gz +0 -0
- package/dist/dashboard/assets/routing-CO9zCUus.js +0 -1
- package/dist/dashboard/assets/routing-CO9zCUus.js.gz +0 -0
- package/dist/dashboard/assets/select-BIYIBkMU.js.gz +0 -0
- package/dist/dashboard/assets/settings-jS3-PNSA.js +0 -1
- package/dist/dashboard/assets/settings-jS3-PNSA.js.gz +0 -0
- package/dist/dashboard/assets/stack-DfgnFDWY.js.gz +0 -0
- package/dist/dashboard/assets/state-CkfdYWel.js.gz +0 -0
- package/dist/dashboard/assets/stream-retry-5QnVuSSo.js +0 -1
- package/dist/dashboard/assets/stream-retry-5QnVuSSo.js.gz +0 -0
- package/dist/dashboard/assets/switch-SYm237HG.js.gz +0 -0
- package/dist/dashboard/assets/useMutation-BQqCRNIi.js.gz +0 -0
- package/src/console/cli-tools/injectors/cline.ts +0 -110
- package/src/console/cli-tools/injectors/copilot.ts +0 -108
- package/src/console/cli-tools/injectors/cowork.ts +0 -64
- package/src/console/cli-tools/injectors/deepseek-tui.ts +0 -85
- package/src/console/cli-tools/injectors/droid.ts +0 -96
- package/src/console/cli-tools/injectors/grok-build.ts +0 -70
- package/src/console/cli-tools/injectors/hermes.ts +0 -101
- package/src/console/cli-tools/injectors/jcode.ts +0 -115
- package/src/console/cli-tools/injectors/openclaw.ts +0 -161
- package/src/network/response-headers.ts +0 -39
- package/src/protocol/response/dsml.ts +0 -38
- package/src/providers/integrations/xiaomi-mimo/think-stream.ts +0 -229
- /package/dist/dashboard/assets/brand/{og_bansos.webp → og_bansos_initial.webp} +0 -0
- /package/dist/dashboard/assets/brand/{og_image.webp → og_image_initial.webp} +0 -0
|
@@ -1,8 +1,35 @@
|
|
|
1
1
|
// Tenant payload-capture policy resolved once and shared by dispatch attempts.
|
|
2
2
|
import type { CartethyiaDatabase } from "../../persistence/db-handle";
|
|
3
|
-
import {
|
|
3
|
+
import { resolvePayloadCapture } from "../../observability/payload-capture";
|
|
4
4
|
import type { TelemetryPayloadDepth, TelemetryPayloadMode } from "../../console/settings/contracts";
|
|
5
5
|
|
|
6
|
+
/** The resolved payload-capture preference, memoized on the request state. */
|
|
7
|
+
export type PayloadCapturePreference =
|
|
8
|
+
| {
|
|
9
|
+
readonly ok: true;
|
|
10
|
+
readonly mode: TelemetryPayloadMode;
|
|
11
|
+
readonly depth: TelemetryPayloadDepth;
|
|
12
|
+
readonly retentionMs: number;
|
|
13
|
+
}
|
|
14
|
+
| { readonly ok: false };
|
|
15
|
+
|
|
16
|
+
/** The slice of request state that memoizes the capture preference. */
|
|
17
|
+
export type CapturePreferenceState = { payloadCapturePreference?: PayloadCapturePreference };
|
|
18
|
+
|
|
19
|
+
/** Resolves the tenant's capture preference once, memoized on the request state. */
|
|
20
|
+
async function resolveCapturePreference(
|
|
21
|
+
db: CartethyiaDatabase,
|
|
22
|
+
tenantId: string | null,
|
|
23
|
+
state: CapturePreferenceState,
|
|
24
|
+
): Promise<PayloadCapturePreference> {
|
|
25
|
+
let preference = state.payloadCapturePreference;
|
|
26
|
+
if (preference === undefined) {
|
|
27
|
+
preference = await resolvePayloadCapture(db, tenantId);
|
|
28
|
+
state.payloadCapturePreference = preference;
|
|
29
|
+
}
|
|
30
|
+
return preference;
|
|
31
|
+
}
|
|
32
|
+
|
|
6
33
|
/** Per-stage body cap for the provider request/response tee. */
|
|
7
34
|
export function providerCaptureMaxBytes(depth?: TelemetryPayloadDepth): number {
|
|
8
35
|
if (depth === "moderate") return 2 * 1024 * 1024;
|
|
@@ -14,14 +41,9 @@ export function providerCaptureMaxBytes(depth?: TelemetryPayloadDepth): number {
|
|
|
14
41
|
export async function shouldCaptureProviderBodyForRequest(
|
|
15
42
|
db: CartethyiaDatabase,
|
|
16
43
|
tenantId: string | null,
|
|
17
|
-
state:
|
|
44
|
+
state: CapturePreferenceState,
|
|
18
45
|
): Promise<boolean> {
|
|
19
|
-
|
|
20
|
-
state.payloadCapturePreference = await resolvePayloadCapture(db, tenantId);
|
|
21
|
-
}
|
|
22
|
-
const resolved = state.payloadCapturePreference as
|
|
23
|
-
| { readonly ok: true; readonly mode: string }
|
|
24
|
-
| { readonly ok: false };
|
|
46
|
+
const resolved = await resolveCapturePreference(db, tenantId, state);
|
|
25
47
|
return resolved.ok === true && resolved.mode === "full";
|
|
26
48
|
}
|
|
27
49
|
|
|
@@ -29,14 +51,9 @@ export async function shouldCaptureProviderBodyForRequest(
|
|
|
29
51
|
export async function resolveCaptureModeForRequest(
|
|
30
52
|
db: CartethyiaDatabase,
|
|
31
53
|
tenantId: string | null,
|
|
32
|
-
state:
|
|
54
|
+
state: CapturePreferenceState,
|
|
33
55
|
): Promise<TelemetryPayloadMode> {
|
|
34
|
-
|
|
35
|
-
state.payloadCapturePreference = await resolvePayloadCapture(db, tenantId);
|
|
36
|
-
}
|
|
37
|
-
const resolved = state.payloadCapturePreference as
|
|
38
|
-
| { readonly ok: true; readonly mode: TelemetryPayloadMode }
|
|
39
|
-
| { readonly ok: false };
|
|
56
|
+
const resolved = await resolveCapturePreference(db, tenantId, state);
|
|
40
57
|
return resolved.ok === true ? resolved.mode : "none";
|
|
41
58
|
}
|
|
42
59
|
|
|
@@ -44,32 +61,12 @@ export async function resolveCaptureModeForRequest(
|
|
|
44
61
|
export async function resolveCaptureDepthForRequest(
|
|
45
62
|
db: CartethyiaDatabase,
|
|
46
63
|
tenantId: string | null,
|
|
47
|
-
state:
|
|
64
|
+
state: CapturePreferenceState,
|
|
48
65
|
): Promise<TelemetryPayloadDepth> {
|
|
49
|
-
|
|
50
|
-
state.payloadCapturePreference = await resolvePayloadCapture(db, tenantId);
|
|
51
|
-
}
|
|
52
|
-
const resolved = state.payloadCapturePreference as
|
|
53
|
-
| { readonly ok: true; readonly depth: TelemetryPayloadDepth }
|
|
54
|
-
| { readonly ok: false };
|
|
66
|
+
const resolved = await resolveCapturePreference(db, tenantId, state);
|
|
55
67
|
return resolved.ok === true ? resolved.depth : "minimum";
|
|
56
68
|
}
|
|
57
69
|
|
|
58
|
-
/** Resolve the tenant's payload retention once per request. */
|
|
59
|
-
export async function resolveCaptureRetentionForRequest(
|
|
60
|
-
db: CartethyiaDatabase,
|
|
61
|
-
tenantId: string | null,
|
|
62
|
-
state: { payloadCapturePreference?: unknown },
|
|
63
|
-
): Promise<number> {
|
|
64
|
-
if (state.payloadCapturePreference === undefined) {
|
|
65
|
-
state.payloadCapturePreference = await resolvePayloadCapture(db, tenantId);
|
|
66
|
-
}
|
|
67
|
-
const resolved = state.payloadCapturePreference as
|
|
68
|
-
| { readonly ok: true; readonly retentionMs: number }
|
|
69
|
-
| { readonly ok: false };
|
|
70
|
-
return resolved.ok === true ? resolved.retentionMs : payloadRetentionMs();
|
|
71
|
-
}
|
|
72
|
-
|
|
73
70
|
/** Combined body cap for the four payload panels at one configured depth. */
|
|
74
71
|
export function captureDepthMaxBytes(depth: TelemetryPayloadDepth): number {
|
|
75
72
|
if (depth === "moderate") return 8 * 1024 * 1024;
|
|
@@ -52,13 +52,13 @@ export type EmptyStreamFetch = (
|
|
|
52
52
|
* unjittered, so a provider returning 200-with-no-events could drive an
|
|
53
53
|
* unbounded wait, and concurrent clients could resynchronize their retries.
|
|
54
54
|
*/
|
|
55
|
-
|
|
55
|
+
interface EmptyStreamPolicy {
|
|
56
56
|
readonly maxEmptyReplays?: number;
|
|
57
57
|
readonly firstEventTimeoutMs?: number;
|
|
58
58
|
readonly sleep?: (ms: number, signal: AbortSignal) => Promise<void>;
|
|
59
59
|
}
|
|
60
60
|
|
|
61
|
-
|
|
61
|
+
interface EmptyStreamResult {
|
|
62
62
|
readonly response: Response;
|
|
63
63
|
readonly tally: StreamTally;
|
|
64
64
|
readonly endpoint: string;
|
|
@@ -112,8 +112,7 @@ export async function fetchWithEmptyStreamRecovery(
|
|
|
112
112
|
|
|
113
113
|
for (let endpointIndex = 0; endpointIndex < endpoints.length; endpointIndex++) {
|
|
114
114
|
const endpoint = endpoints[endpointIndex]!;
|
|
115
|
-
const
|
|
116
|
-
const replaysAllowed = isLastEndpoint ? maxReplays : maxReplays;
|
|
115
|
+
const replaysAllowed = maxReplays;
|
|
117
116
|
|
|
118
117
|
for (let replay = 0; replay <= replaysAllowed; replay++) {
|
|
119
118
|
// Check abort before each fetch
|
|
@@ -167,12 +166,7 @@ export async function fetchWithEmptyStreamRecovery(
|
|
|
167
166
|
// Empty — replay if we haven't exhausted this endpoint's budget
|
|
168
167
|
if (replay < replaysAllowed) {
|
|
169
168
|
const delay = fallbackRetryDelayMs(replay);
|
|
170
|
-
|
|
171
|
-
await sleep(delay, deps.signal);
|
|
172
|
-
} catch (error) {
|
|
173
|
-
if (error instanceof GatewayError) throw error;
|
|
174
|
-
throw error;
|
|
175
|
-
}
|
|
169
|
+
await sleep(delay, deps.signal);
|
|
176
170
|
}
|
|
177
171
|
}
|
|
178
172
|
}
|
|
@@ -196,9 +190,22 @@ async function fetchWithTimeout(
|
|
|
196
190
|
signal: AbortSignal,
|
|
197
191
|
timeoutMs: number,
|
|
198
192
|
): Promise<Response> {
|
|
199
|
-
// Race the fetch against a first-event timeout
|
|
193
|
+
// Race the fetch against a first-event timeout. The timer and the abort
|
|
194
|
+
// listener are both cleaned up once the race settles, so a successful fetch
|
|
195
|
+
// leaves neither behind.
|
|
196
|
+
let timer: ReturnType<typeof setTimeout> | undefined;
|
|
197
|
+
const onAbort = (): void => {
|
|
198
|
+
if (timer !== undefined) clearTimeout(timer);
|
|
199
|
+
// A client cancellation is the caller's signal; the fetch already races the
|
|
200
|
+
// same abort signal, so rejecting here only surfaces the 499 envelope.
|
|
201
|
+
timeoutPromiseReject(
|
|
202
|
+
new GatewayError("transport_closed", 499, "The request was aborted (request {requestId}). Send it again whenever ready 🙃"),
|
|
203
|
+
);
|
|
204
|
+
};
|
|
205
|
+
let timeoutPromiseReject!: (reason: unknown) => void;
|
|
200
206
|
const timeoutPromise = new Promise<never>((_, reject) => {
|
|
201
|
-
|
|
207
|
+
timeoutPromiseReject = reject;
|
|
208
|
+
timer = setTimeout(() => {
|
|
202
209
|
// A first-event timeout is the gateway waiting on a silent upstream (e.g.
|
|
203
210
|
// a provider that holds the request in a queue), not the client leaving.
|
|
204
211
|
// It must surface as a deadline failure, never as a `transport_closed`
|
|
@@ -206,19 +213,15 @@ async function fetchWithTimeout(
|
|
|
206
213
|
// otherwise hide a real upstream wait behind "cancelled by client".
|
|
207
214
|
reject(new GatewayError("deadline_exceeded", 504, "First event timeout (request {requestId}). Some things just take longer 🐢", {}, "upstream"));
|
|
208
215
|
}, timeoutMs);
|
|
209
|
-
signal.addEventListener("abort", () => {
|
|
210
|
-
clearTimeout(timer);
|
|
211
|
-
reject(new GatewayError("transport_closed", 499, "The request was aborted (request {requestId}). Send it again whenever ready 🙃"));
|
|
212
|
-
}, { once: true });
|
|
213
216
|
});
|
|
214
|
-
|
|
217
|
+
signal.addEventListener("abort", onAbort, { once: true });
|
|
215
218
|
try {
|
|
216
219
|
return await Promise.race([
|
|
217
220
|
fetch(url, { headers: request.headers, body: request.body, signal }),
|
|
218
221
|
timeoutPromise,
|
|
219
222
|
]);
|
|
220
|
-
}
|
|
221
|
-
if (
|
|
222
|
-
|
|
223
|
+
} finally {
|
|
224
|
+
if (timer !== undefined) clearTimeout(timer);
|
|
225
|
+
signal.removeEventListener("abort", onAbort);
|
|
223
226
|
}
|
|
224
227
|
}
|
|
@@ -20,20 +20,26 @@
|
|
|
20
20
|
*/
|
|
21
21
|
import type { ImageGenerationOutcome, ProviderAdapter, ProviderDispatchTarget } from "../../providers/provider-contracts";
|
|
22
22
|
import { GatewayError } from "../../protocol/gateway-error";
|
|
23
|
-
import { resolveCredentialForAccount } from "../../providers/operations/provider-credential-service";
|
|
24
23
|
import type { ServiceKind } from "../../protocol/canonical-model";
|
|
25
24
|
import { metrics } from "../../observability/metrics";
|
|
26
25
|
import type { ProxyRequestState } from "../request/state";
|
|
27
26
|
import { completeAttempt, estimatedUsage, nativeCompletionBase } from "./attempt-finalize";
|
|
28
27
|
import { runAttemptLoop } from "./attempt-loop";
|
|
29
28
|
import { captureProviderRateLimit } from "./ratelimit-capture";
|
|
29
|
+
import {
|
|
30
|
+
hasScope,
|
|
31
|
+
jsonResponse,
|
|
32
|
+
networkFetchContext,
|
|
33
|
+
requireRecordBody,
|
|
34
|
+
resolveAttemptCredential,
|
|
35
|
+
} from "./native-shared";
|
|
30
36
|
|
|
31
37
|
import type { GatewayHandlerDeps } from "./handler-deps";
|
|
32
38
|
|
|
33
|
-
|
|
39
|
+
type GenerationHandlerDeps = GatewayHandlerDeps;
|
|
34
40
|
|
|
35
41
|
/** Which surface a handler serves. Drives candidate filtering and error copy. */
|
|
36
|
-
|
|
42
|
+
type GenerationKind = Extract<ServiceKind, "imageGeneration" | "videoGeneration">;
|
|
37
43
|
|
|
38
44
|
/**
|
|
39
45
|
* Surfaces the gateway routes but does not model.
|
|
@@ -44,7 +50,7 @@ export type GenerationKind = Extract<ServiceKind, "imageGeneration" | "videoGene
|
|
|
44
50
|
* is the whole feature: routing, admission, failover, and accounting still
|
|
45
51
|
* apply, and the upstream's own bytes come back untouched.
|
|
46
52
|
*/
|
|
47
|
-
|
|
53
|
+
type PassthroughKind = Extract<ServiceKind, "embedding" | "audioTranscription" | "audioSpeech">;
|
|
48
54
|
|
|
49
55
|
/** Error-copy noun per surface, so a refusal names what the caller asked for. */
|
|
50
56
|
const SURFACE_NOUNS: Record<GenerationKind | PassthroughKind, string> = {
|
|
@@ -74,13 +80,6 @@ function isMediaKind(kind: GenerationKind | PassthroughKind): kind is Generation
|
|
|
74
80
|
*/
|
|
75
81
|
const MAX_ITEMS_PER_REQUEST = 10;
|
|
76
82
|
|
|
77
|
-
function jsonResponse(payload: unknown, status: number, requestId: string): Response {
|
|
78
|
-
return new Response(JSON.stringify(payload), {
|
|
79
|
-
status,
|
|
80
|
-
headers: { "content-type": "application/json", "x-request-id": requestId },
|
|
81
|
-
});
|
|
82
|
-
}
|
|
83
|
-
|
|
84
83
|
/**
|
|
85
84
|
* Forwards a streaming upstream response to the client and accounts for it.
|
|
86
85
|
*
|
|
@@ -145,12 +144,12 @@ async function passThrough(
|
|
|
145
144
|
}
|
|
146
145
|
|
|
147
146
|
/** Builds the handler for one generation surface. */
|
|
148
|
-
|
|
147
|
+
function createGenerationHandler(kind: GenerationKind, deps: GenerationHandlerDeps) {
|
|
149
148
|
return createNativeSurfaceHandler(kind, deps);
|
|
150
149
|
}
|
|
151
150
|
|
|
152
151
|
/** Builds the handler for one routed surface, media or passthrough. */
|
|
153
|
-
|
|
152
|
+
function createNativeSurfaceHandler(
|
|
154
153
|
kind: GenerationKind | PassthroughKind,
|
|
155
154
|
deps: GenerationHandlerDeps,
|
|
156
155
|
) {
|
|
@@ -158,19 +157,14 @@ export function createNativeSurfaceHandler(
|
|
|
158
157
|
return async ({ request }: { request: Request }): Promise<Response> => {
|
|
159
158
|
const state = deps.stateStore.require(request);
|
|
160
159
|
const authorization = state.authorization;
|
|
161
|
-
const
|
|
162
|
-
|
|
163
|
-
|
|
164
|
-
|
|
165
|
-
400,
|
|
166
|
-
`The ${noun} body must be a JSON object (request {requestId}). Send one with at least a model and a prompt. Check the shape of what you sent — then resend 🪞`,
|
|
167
|
-
);
|
|
168
|
-
}
|
|
169
|
-
const generation = body as Record<string, unknown>;
|
|
160
|
+
const generation = requireRecordBody(
|
|
161
|
+
state,
|
|
162
|
+
`The ${noun} body must be a JSON object (request {requestId}). Send one with at least a model and a prompt. Check the shape of what you sent — then resend 🪞`,
|
|
163
|
+
);
|
|
170
164
|
if (typeof generation.model !== "string" || generation.model.trim().length === 0) {
|
|
171
165
|
throw new GatewayError("invalid_request", 400, `A model is required (request {requestId}): it names the ${noun} provider to use. Check the shape of what you sent — then resend 🪞`);
|
|
172
166
|
}
|
|
173
|
-
if (!authorization
|
|
167
|
+
if (!hasScope(authorization?.snapshot.scopes, "routing:invoke")) {
|
|
174
168
|
throw new GatewayError(
|
|
175
169
|
"policy_rejected",
|
|
176
170
|
403,
|
|
@@ -200,10 +194,15 @@ export function createNativeSurfaceHandler(
|
|
|
200
194
|
const prepared = await deps.proxyPreparer.prepareNativeService({
|
|
201
195
|
model: generation.model,
|
|
202
196
|
serviceKind: kind,
|
|
203
|
-
authorization
|
|
197
|
+
// `requireRecordBody` already refused a missing authorization; this guard
|
|
198
|
+
// exists only to narrow the type for the code below.
|
|
199
|
+
authorization: authorization!,
|
|
204
200
|
...(state.abortController.signal ? { signal: state.abortController.signal } : {}),
|
|
205
201
|
...(state.clientUserAgent === undefined ? {} : { clientUserAgent: state.clientUserAgent }),
|
|
206
202
|
});
|
|
203
|
+
if (!authorization) {
|
|
204
|
+
throw new GatewayError("invalid_request", 400, "authentication required (request {requestId})");
|
|
205
|
+
}
|
|
207
206
|
|
|
208
207
|
return runAttemptLoop<Response, ProviderAdapter>({
|
|
209
208
|
state,
|
|
@@ -221,19 +220,7 @@ export function createNativeSurfaceHandler(
|
|
|
221
220
|
strictPoolSelection: false,
|
|
222
221
|
exhaustedError: new GatewayError("route_unavailable", 503, `no eligible ${noun} generation route (request {requestId}). The bouncer stepped out — come back shortly 🚪`),
|
|
223
222
|
prepare: async (candidate) => {
|
|
224
|
-
const credential = candidate
|
|
225
|
-
? await resolveCredentialForAccount(
|
|
226
|
-
deps.db,
|
|
227
|
-
candidate.provider_id,
|
|
228
|
-
candidate.provider_account_id,
|
|
229
|
-
deps.oauthRefreshService && deps.resolveOAuthRefresher
|
|
230
|
-
? { refreshService: deps.oauthRefreshService, resolveRefresher: deps.resolveOAuthRefresher }
|
|
231
|
-
: undefined,
|
|
232
|
-
)
|
|
233
|
-
: {
|
|
234
|
-
provider_id: candidate.provider_id as ProviderAdapter["provider_id"],
|
|
235
|
-
credential_kind: "none" as const,
|
|
236
|
-
};
|
|
223
|
+
const credential = await resolveAttemptCredential(deps, candidate);
|
|
237
224
|
const adapter = await deps.resolveProviderAdapter?.(candidate.provider_id);
|
|
238
225
|
if (!adapter) {
|
|
239
226
|
throw new GatewayError(
|
|
@@ -309,14 +296,7 @@ export function createNativeSurfaceHandler(
|
|
|
309
296
|
abort_signal: state.abortController.signal,
|
|
310
297
|
request_headers: inboundHeaders,
|
|
311
298
|
...(candidate.user_agent === undefined ? {} : { user_agent: candidate.user_agent }),
|
|
312
|
-
...(deps.
|
|
313
|
-
? {
|
|
314
|
-
outbound_fetch: deps.networkBindingFactory.fetch(
|
|
315
|
-
leases.proxySlot?.poolId,
|
|
316
|
-
authorization.tenantId,
|
|
317
|
-
),
|
|
318
|
-
}
|
|
319
|
-
: {}),
|
|
299
|
+
...networkFetchContext(deps, leases.proxySlot?.poolId, authorization.tenantId),
|
|
320
300
|
};
|
|
321
301
|
// Prefer the provider's own method for this kind; otherwise post the
|
|
322
302
|
// body to the candidate's endpoint through the adapter, which owns the
|
|
@@ -0,0 +1,127 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* Shared helpers for the native dispatch handlers (web search, token count,
|
|
3
|
+
* System One, responses-compact, generation). The handlers previously spelled
|
|
4
|
+
* the same credential-resolution, outbound-fetch, body-validation, thinking-
|
|
5
|
+
* suffix, response-serialization, and usage-clone blocks by hand, and the
|
|
6
|
+
* copies could drift (generation's `jsonResponse` lost the `cache-control`
|
|
7
|
+
* header). One home here keeps them in step.
|
|
8
|
+
*/
|
|
9
|
+
import { GatewayError } from "../../protocol/gateway-error";
|
|
10
|
+
import type { ProviderId, ResolvedCredential, ValidatedOutboundFetch } from "../../providers/provider-contracts";
|
|
11
|
+
import { resolveCredentialForAccount } from "../../providers/operations/provider-credential-service";
|
|
12
|
+
import { repriceUsage, usageFromProvider } from "../../providers/provider-usage";
|
|
13
|
+
import type { UsageRecord } from "../../protocol/canonical-model";
|
|
14
|
+
import { parseThinkingSuffix } from "../../protocol/translation/thinking";
|
|
15
|
+
import type { GatewayHandlerDeps } from "./handler-deps";
|
|
16
|
+
|
|
17
|
+
/**
|
|
18
|
+
* Whether a scope list contains the named scope. The native handlers used to
|
|
19
|
+
* force `scopes?.includes(name as never)` — a type-loop hole that bypassed
|
|
20
|
+
* the `AccessScope` union only to compare runtime strings. The check is the
|
|
21
|
+
* same, with a real `readonly string[]` receiver.
|
|
22
|
+
*/
|
|
23
|
+
export function hasScope(scopes: readonly string[] | undefined, name: string): boolean {
|
|
24
|
+
return scopes?.includes(name) === true;
|
|
25
|
+
}
|
|
26
|
+
|
|
27
|
+
/**
|
|
28
|
+
* Resolves the credential for one candidate: a real account credential when
|
|
29
|
+
* the candidate carries an account, otherwise the credential-less `"none"`
|
|
30
|
+
* credential. Shared by every native handler's `prepare`.
|
|
31
|
+
*/
|
|
32
|
+
export async function resolveAttemptCredential(
|
|
33
|
+
deps: Pick<GatewayHandlerDeps, "db" | "oauthRefreshService" | "resolveOAuthRefresher">,
|
|
34
|
+
candidate: { readonly provider_id: string; readonly provider_account_id?: string },
|
|
35
|
+
): Promise<ResolvedCredential> {
|
|
36
|
+
if (!candidate.provider_account_id) {
|
|
37
|
+
return {
|
|
38
|
+
provider_id: candidate.provider_id as ProviderId,
|
|
39
|
+
credential_kind: "none" as const,
|
|
40
|
+
};
|
|
41
|
+
}
|
|
42
|
+
return resolveCredentialForAccount(
|
|
43
|
+
deps.db,
|
|
44
|
+
candidate.provider_id,
|
|
45
|
+
candidate.provider_account_id,
|
|
46
|
+
deps.oauthRefreshService && deps.resolveOAuthRefresher
|
|
47
|
+
? { refreshService: deps.oauthRefreshService, resolveRefresher: deps.resolveOAuthRefresher }
|
|
48
|
+
: undefined,
|
|
49
|
+
);
|
|
50
|
+
}
|
|
51
|
+
|
|
52
|
+
/** The outbound-fetch context fragment native handlers spread into the dispatch context. */
|
|
53
|
+
export function networkFetchContext(
|
|
54
|
+
deps: { readonly networkBindingFactory?: GatewayHandlerDeps["networkBindingFactory"] },
|
|
55
|
+
poolId: string | undefined,
|
|
56
|
+
tenantId: string | null,
|
|
57
|
+
): { outbound_fetch?: ValidatedOutboundFetch } {
|
|
58
|
+
if (!deps.networkBindingFactory) return {};
|
|
59
|
+
return { outbound_fetch: deps.networkBindingFactory.fetch(poolId, tenantId ?? undefined) };
|
|
60
|
+
}
|
|
61
|
+
|
|
62
|
+
/** Serializes a JSON payload with the shared native-response header set. */
|
|
63
|
+
export function jsonResponse(payload: unknown, status: number, requestId: string): Response {
|
|
64
|
+
return new Response(JSON.stringify(payload), {
|
|
65
|
+
status,
|
|
66
|
+
headers: {
|
|
67
|
+
"content-type": "application/json",
|
|
68
|
+
"cache-control": "no-store",
|
|
69
|
+
"x-request-id": requestId,
|
|
70
|
+
},
|
|
71
|
+
});
|
|
72
|
+
}
|
|
73
|
+
|
|
74
|
+
/**
|
|
75
|
+
* Strips a thinking suffix (`model(high)`) from the model name and rebuilds
|
|
76
|
+
* the body with the bare model when it changed. Returns the bare model and
|
|
77
|
+
* the body to forward upstream.
|
|
78
|
+
*/
|
|
79
|
+
export function stripThinkingSuffix(
|
|
80
|
+
model: string,
|
|
81
|
+
body: Record<string, unknown>,
|
|
82
|
+
): { bareModel: string; body: Record<string, unknown> } {
|
|
83
|
+
const { model: bareModel } = parseThinkingSuffix(model);
|
|
84
|
+
return { bareModel, body: bareModel === model ? body : { ...body, model: bareModel } };
|
|
85
|
+
}
|
|
86
|
+
|
|
87
|
+
/**
|
|
88
|
+
* Validates the ingress body is a non-null object and returns it as a record.
|
|
89
|
+
* Throws the standard `invalid_request` with the caller's message (kept
|
|
90
|
+
* per-route so the wire text is unchanged).
|
|
91
|
+
*/
|
|
92
|
+
export function requireRecordBody(
|
|
93
|
+
state: { readonly authorization?: unknown; readonly ingressBody?: unknown },
|
|
94
|
+
message: string,
|
|
95
|
+
): Record<string, unknown> {
|
|
96
|
+
const { authorization, ingressBody } = state;
|
|
97
|
+
if (!authorization || ingressBody === null || typeof ingressBody !== "object" || Array.isArray(ingressBody))
|
|
98
|
+
throw new GatewayError("invalid_request", 400, message);
|
|
99
|
+
return ingressBody as Record<string, unknown>;
|
|
100
|
+
}
|
|
101
|
+
|
|
102
|
+
/**
|
|
103
|
+
* Reads usage from an upstream response clone, repricing the reported value (or
|
|
104
|
+
* the caller's estimate) for the routed provider/model. Shared by the native
|
|
105
|
+
* handlers that proxy an upstream JSON body verbatim.
|
|
106
|
+
*/
|
|
107
|
+
export async function usageFromUpstreamClone(
|
|
108
|
+
response: Response,
|
|
109
|
+
estimate: UsageRecord,
|
|
110
|
+
providerId: string,
|
|
111
|
+
modelId: string,
|
|
112
|
+
): Promise<UsageRecord> {
|
|
113
|
+
let reportedUsage: ReturnType<typeof usageFromProvider>;
|
|
114
|
+
try {
|
|
115
|
+
const parsed: unknown = await response.clone().json();
|
|
116
|
+
reportedUsage = usageFromProvider(
|
|
117
|
+
typeof parsed === "object" && parsed !== null
|
|
118
|
+
? (parsed as Record<string, unknown>)["usage"]
|
|
119
|
+
: undefined,
|
|
120
|
+
);
|
|
121
|
+
} catch {
|
|
122
|
+
reportedUsage = undefined;
|
|
123
|
+
}
|
|
124
|
+
return repriceUsage(reportedUsage ?? estimate, providerId, modelId);
|
|
125
|
+
}
|
|
126
|
+
|
|
127
|
+
|
|
@@ -41,7 +41,7 @@ import type { PreparedProxyRequest } from "../request/preparer";
|
|
|
41
41
|
|
|
42
42
|
import type { GatewayHandlerDeps } from "./handler-deps";
|
|
43
43
|
|
|
44
|
-
|
|
44
|
+
type ProviderProxyHandlerDeps = GatewayHandlerDeps & {
|
|
45
45
|
/** Live lookup of a provider's SSRF-validated upstream host. */
|
|
46
46
|
readonly byokUpstreamHosts?: { readonly get: (providerId: string) => ByokUpstreamHost | undefined };
|
|
47
47
|
};
|
|
@@ -135,6 +135,7 @@ export async function handleProviderProxyRequest(
|
|
|
135
135
|
{
|
|
136
136
|
providerId: candidateForPrompt.provider_id,
|
|
137
137
|
modelId: candidateForPrompt.model_id,
|
|
138
|
+
requestedModel: prepared.plan?.requested_model ?? candidateForPrompt.model_id,
|
|
138
139
|
},
|
|
139
140
|
)?.text
|
|
140
141
|
: undefined;
|
|
@@ -239,6 +240,7 @@ export async function handleProviderProxyRequest(
|
|
|
239
240
|
...(candidate.output_limit == null ? {} : { output_limit: candidate.output_limit }),
|
|
240
241
|
capabilities: candidate.capability_profile,
|
|
241
242
|
user_agent: candidate.user_agent,
|
|
243
|
+
...(candidate.dialect === undefined ? {} : { dialect: candidate.dialect }),
|
|
242
244
|
};
|
|
243
245
|
const candidateRequest =
|
|
244
246
|
candidate.model_id === dispatchRequest.model
|
|
@@ -1,11 +1,15 @@
|
|
|
1
1
|
import { GatewayError } from "../../protocol/gateway-error";
|
|
2
|
-
import { resolveCredentialForAccount } from "../../providers/operations/provider-credential-service";
|
|
3
2
|
import { metrics } from "../../observability/metrics";
|
|
4
|
-
import { parseThinkingSuffix } from "../../protocol/translation/thinking";
|
|
5
3
|
import type { CodexCompactAdapter } from "../../providers/integrations/codex/codex";
|
|
6
4
|
import { completeAttempt, estimatedUsage, nativeCompletionBase } from "./attempt-finalize";
|
|
7
|
-
import { repriceUsage, usageFromProvider } from "../../providers/provider-usage";
|
|
8
5
|
import { runAttemptLoop } from "./attempt-loop";
|
|
6
|
+
import {
|
|
7
|
+
networkFetchContext,
|
|
8
|
+
requireRecordBody,
|
|
9
|
+
resolveAttemptCredential,
|
|
10
|
+
stripThinkingSuffix,
|
|
11
|
+
usageFromUpstreamClone,
|
|
12
|
+
} from "./native-shared";
|
|
9
13
|
|
|
10
14
|
/**
|
|
11
15
|
* The native Codex compaction route (`POST /v1/responses/compact`) and the
|
|
@@ -13,7 +17,7 @@ import { runAttemptLoop } from "./attempt-loop";
|
|
|
13
17
|
*/
|
|
14
18
|
import type { GatewayHandlerDeps } from "./handler-deps";
|
|
15
19
|
|
|
16
|
-
|
|
20
|
+
type ResponsesCompactHandlerDeps = GatewayHandlerDeps;
|
|
17
21
|
|
|
18
22
|
/**
|
|
19
23
|
* Handles `POST /v1/responses/compact`: the native Codex compaction proxy. The
|
|
@@ -25,10 +29,10 @@ export function createResponsesCompactHandler(deps: ResponsesCompactHandlerDeps)
|
|
|
25
29
|
return async ({ request }: { request: Request }): Promise<Response> => {
|
|
26
30
|
const state = deps.stateStore.require(request);
|
|
27
31
|
const authorization = state.authorization;
|
|
28
|
-
const
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
32
|
+
const compact = requireRecordBody(
|
|
33
|
+
state,
|
|
34
|
+
"compact request body must be a JSON object (request {requestId}). Check the shape of what you sent — then resend 🪞",
|
|
35
|
+
);
|
|
32
36
|
if (typeof compact.model !== "string" || compact.model.trim().length === 0)
|
|
33
37
|
throw new GatewayError("invalid_request", 400, "model is required (request {requestId}). Check the shape of what you sent — then resend 🪞");
|
|
34
38
|
// A thinking suffix (`model(high)`) is stripped here for the same reason the
|
|
@@ -39,9 +43,7 @@ export function createResponsesCompactHandler(deps: ResponsesCompactHandlerDeps)
|
|
|
39
43
|
// reasoning effort to shape, and the body is passed upstream verbatim.
|
|
40
44
|
// Stripping keeps a client that appends the suffix to every request working
|
|
41
45
|
// here instead of failing with a misleading model-not-found.
|
|
42
|
-
const {
|
|
43
|
-
const compactBody: Record<string, unknown> =
|
|
44
|
-
compactModel === compact.model ? compact : { ...compact, model: compactModel };
|
|
46
|
+
const { bareModel: compactModel, body: compactBody } = stripThinkingSuffix(compact.model, compact);
|
|
45
47
|
if (!(typeof compact.input === "string" || Array.isArray(compact.input)))
|
|
46
48
|
throw new GatewayError("invalid_request", 400, "input must be a string or array (request {requestId}). Check the shape of what you sent — then resend 🪞");
|
|
47
49
|
if ("instructions" in compact && typeof compact.instructions !== "string")
|
|
@@ -57,7 +59,7 @@ export function createResponsesCompactHandler(deps: ResponsesCompactHandlerDeps)
|
|
|
57
59
|
throw new GatewayError("admission_unavailable", 503, "Codex compact transport unavailable (request {requestId}). The bouncer stepped out — come back shortly 🚪");
|
|
58
60
|
const prepared = await deps.proxyPreparer.prepareNativeCompact({
|
|
59
61
|
model: compactModel,
|
|
60
|
-
authorization
|
|
62
|
+
authorization: authorization!,
|
|
61
63
|
signal: state.abortController.signal,
|
|
62
64
|
...(state.clientUserAgent === undefined ? {} : { clientUserAgent: state.clientUserAgent }),
|
|
63
65
|
});
|
|
@@ -88,14 +90,10 @@ export function createResponsesCompactHandler(deps: ResponsesCompactHandlerDeps)
|
|
|
88
90
|
503,
|
|
89
91
|
"Codex compact requires an OAuth account (request {requestId}). Wrong key, right door — check the credential 🔑",
|
|
90
92
|
);
|
|
91
|
-
const credential = await
|
|
92
|
-
|
|
93
|
-
"codex",
|
|
94
|
-
|
|
95
|
-
deps.oauthRefreshService && deps.resolveOAuthRefresher
|
|
96
|
-
? { refreshService: deps.oauthRefreshService, resolveRefresher: deps.resolveOAuthRefresher }
|
|
97
|
-
: undefined,
|
|
98
|
-
);
|
|
93
|
+
const credential = await resolveAttemptCredential(deps, {
|
|
94
|
+
...candidate,
|
|
95
|
+
provider_id: "codex",
|
|
96
|
+
});
|
|
99
97
|
if (credential.credential_kind !== "oauth")
|
|
100
98
|
throw new GatewayError(
|
|
101
99
|
"capability_unsupported",
|
|
@@ -110,36 +108,17 @@ export function createResponsesCompactHandler(deps: ResponsesCompactHandlerDeps)
|
|
|
110
108
|
credential,
|
|
111
109
|
deadline: state.deadlineMs,
|
|
112
110
|
abort_signal: state.abortController.signal,
|
|
113
|
-
...(deps.
|
|
114
|
-
? {
|
|
115
|
-
outbound_fetch: deps.networkBindingFactory.fetch(
|
|
116
|
-
leases.proxySlot?.poolId,
|
|
117
|
-
authorization.tenantId,
|
|
118
|
-
),
|
|
119
|
-
}
|
|
120
|
-
: {}),
|
|
111
|
+
...networkFetchContext(deps, leases.proxySlot?.poolId, authorization!.tenantId),
|
|
121
112
|
});
|
|
122
113
|
// Prefer the usage the compaction endpoint actually reported. The
|
|
123
114
|
// estimate is a fixed 1024+1024 reserve, and compaction is by
|
|
124
115
|
// construction a large-input operation, so charging it unconditionally
|
|
125
|
-
// let real spend run far past the counter. `
|
|
126
|
-
//
|
|
127
|
-
//
|
|
128
|
-
|
|
129
|
-
|
|
130
|
-
|
|
131
|
-
const parsed: unknown = await response.clone().json();
|
|
132
|
-
reportedUsage = usageFromProvider(
|
|
133
|
-
typeof parsed === "object" && parsed !== null
|
|
134
|
-
? (parsed as Record<string, unknown>)["usage"]
|
|
135
|
-
: undefined,
|
|
136
|
-
);
|
|
137
|
-
} catch {
|
|
138
|
-
reportedUsage = undefined;
|
|
139
|
-
}
|
|
140
|
-
const usage = repriceUsage(
|
|
141
|
-
reportedUsage ??
|
|
142
|
-
estimatedUsage(prepared.estimatedInputTokens, prepared.estimatedOutputTokens),
|
|
116
|
+
// let real spend run far past the counter. `usageFromUpstreamClone`
|
|
117
|
+
// falls back to the estimate when the body reported nothing, and
|
|
118
|
+
// `response.clone()` leaves the body readable for the client.
|
|
119
|
+
const usage = await usageFromUpstreamClone(
|
|
120
|
+
response,
|
|
121
|
+
estimatedUsage(prepared.estimatedInputTokens, prepared.estimatedOutputTokens),
|
|
143
122
|
candidate.provider_id,
|
|
144
123
|
candidate.model_id,
|
|
145
124
|
);
|