Yichen Jiang 1d2ea70e9b feat(llm-deepseek): per-model output cap in the advisory catalog ...

A catalog entry may now carry its own `maxTokens`, matching the shape the
pi-ai adapter already exposes. Exact-model resolution prefers it over the
profile value, so capping one model no longer means capping the route.

Defaults are unchanged: an entry without a cap, and any unlisted
pass-through id, still resolve to the profile `maxTokens` (256,000), and the
shipped catalog keeps its context windows.

2026-08-04 14:53:02 +08:00

..

adapter.e2e.ts

refactor(llm-deepseek)!: rename the provider route to deepseek-official

2026-07-29 16:36:07 +08:00

adapter.spec.ts

feat(llm-deepseek): per-model output cap in the advisory catalog

2026-08-04 14:53:02 +08:00

assemble.ts

refactor(llm-deepseek)!: rename the provider route to deepseek-official

2026-07-29 16:36:07 +08:00

dynamic-config.spec.ts

fix(web-config): close the wire boundary, the redacted-replace data loss, and three P2s

2026-07-30 18:30:15 +08:00

loader-composition.spec.ts

fix(cli)!: stop hoisting $DSH_HOME/.env into process.env

2026-07-30 15:44:32 +08:00

mock-server.ts

feat(llm-deepseek): per-request connection resolution over settings + credentials

2026-07-29 13:26:52 +08:00

serialize.spec.ts

fix(agent-loop): rematerialize adapter defaults

2026-07-30 21:49:58 +08:00

sse.spec.ts

refactor(llm-deepseek): replace hand-rolled SSE parser with eventsource-parser

2026-07-26 22:45:33 +08:00

translate.spec.ts

fix(llm): classify empty model completions as retryable EMPTY_RESPONSE

2026-07-24 19:23:12 +08:00