History

Yichen Jiang 915a56208d Merge pull request #1977 from deepseek-harness/worktree/pi-ai-per-model-reasoning feat(llm-pi-ai): per-model reasoning efforts, dialect switches, and modelOverrides in settings.yaml		2026-08-08 14:23:07 +08:00
..
llm	cleanup: remove private repository references	2026-08-08 01:10:57 +08:00
llm-deepseek	cleanup(environment): keep one lookup order	2026-08-07 23:33:11 +08:00
llm-pi-ai	docs(llm-pi-ai): state the composition-base assumption for the dict-merge limitation	2026-08-08 11:06:29 +08:00
llm-retry	docs(config): align environment and credential contracts	2026-08-07 22:04:04 +08:00
token-meter	fix: projection error	2026-08-06 19:36:15 +08:00
README.i18n.yaml	docs: rescan rebased documentation hierarchy	2026-08-05 16:16:57 +08:00
README.md	docs: apply hierarchy across the corpus	2026-08-05 16:13:28 +08:00
README.zh.md	docs: rescan rebased documentation hierarchy	2026-08-05 16:16:57 +08:00

README.md

llm/ — LLM capability family

English | 中文

The LLM seam and its provider adapters. The interface package (llm) owns the abstract service, the content-block vocabulary, and the stream-chunk assembler; the adapters are concrete implementations that register on ctx.llm. All product packages.

Package	Role	ctx key
`llm/`	LLM service and shared streaming vocabulary	`ctx.llm`
`token-meter/`	Replay-aware token measurement	`ctx.tokenMeter`
`llm-retry/`	Provider-scoped retry policy	listens to `agent/request-error`
`llm-deepseek/`	Direct DeepSeek adapter	registers on `ctx.llm`
`llm-pi-ai/`	Multi-provider pi-ai adapter	registers on `ctx.llm`

Adapters register provider routes on the seam; retry and token measurement remain separate consumers. The child READMEs own routing, metadata, replay, and provider-wire details; the LLM architecture decisions own the rationale.