Coder/deepseek-harness
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(llm-deepseek): support low reasoning effort

Browse Source
This commit is contained in:
Yichen Jiang
2026-08-14 11:47:59 +08:00
parent f322d5cb0d
commit 226600147e
21 changed files with 54 additions and 46 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml
View File

@@ -2,5 +2,5 @@
# side as of the last confirmed-consistent state. Both languages carry equal authority;
# after editing either side, bring the other along and re-record with:
# pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md
2026-07-24-adapter-owned-reasoning-effort-capabilities.md: f71a15626b97090a89f071e552a12b1f89173620
2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md: 6e8c99f98107e2855aa7cb7ea759544013d0ea8a
2026-07-24-adapter-owned-reasoning-effort-capabilities.md: f59e5d351970e785e6ade32797f1c1a5dd9e2591
2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md: 6e657a90e8ef3cd400a3e252f728bb1aadd9f4f2

2
.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md
View File

@@ -14,7 +14,7 @@ Reasoning strength was adapter configuration only, so a conversation could not d
`LlmCallConfig` and `GenerateOptions` carry the optional effort. The agent loop prepares the post-`agent/request` config under the active turn signal before writing `request/header`, so defaults and dynamic changes are model-visible only after becoming durable facts. The prepared call retains the exact adapter registration across asynchronous exact-model resolution, durable header logging, and dispatch; direct `LlmRuntime.stream()` calls likewise capture their final registration before awaiting resolution. A route with no registered adapter retains its proposed config so an `llm/stream` middleware can own and short-circuit it; terminal dispatch still rejects an unhandled route. A resumed loop retains the logged effort only when its initial provider/model route is unchanged; a route change discards the previous model's opaque id.
The native DeepSeek adapter advertises `off`, `high`, and `max` when deployment policy permits thinking, and defaults to the configured effort or `high`. Its adapter-owned `off` maps to `thinking.type: disabled` with no `reasoning_effort`; `high` and `max` enable thinking and carry their official wire effort. A `thinking: disabled` deployment publishes only `off` and rejects attempts to enable thinking before provider I/O. The pi-ai adapter publishes each exact model's `getSupportedThinkingLevels()` result unchanged, including `off`, preserves an absent profile default as a provider default, and leaves provider wire-value mapping inside pi-ai. Its common stream options represent `off` by omitting `reasoning`, as required by pi-ai's own API.
The native DeepSeek adapter advertises `off`, `low`, `high`, and `max` when deployment policy permits thinking, and defaults to the configured effort or `high`. Its adapter-owned `off` maps to `thinking.type: disabled` with no `reasoning_effort`; `low`, `high`, and `max` enable thinking and carry their same-named official wire effort. A `thinking: disabled` deployment publishes only `off` and rejects attempts to enable thinking before provider I/O. The pi-ai adapter publishes each exact model's `getSupportedThinkingLevels()` result unchanged, including `off`, preserves an absent profile default as a provider default, and leaves provider wire-value mapping inside pi-ai. Its common stream options represent `off` by omitting `reasoning`, as required by pi-ai's own API.
## Alternatives considered

2
.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md
View File

@@ -14,7 +14,7 @@ Status: implemented
`LlmCallConfig` 和 `GenerateOptions` 携带可选的推理强度。agent loop 在活跃轮次信号的控制下准备 `agent/request` 处理完成后的配置，再写入 `request/header`，因此默认值和动态变更只有成为持久化事实后才对模型可见。准备完成的调用在异步确切模型解析、请求头持久记录和分派全程保留同一项确切的适配器注册；直接调用 `LlmRuntime.stream()` 时，也会在等待解析前捕获最终的适配器注册。没有已注册适配器的路由会保留原定配置，使 `llm/stream` 中间件可以接管并短路该请求；若仍未得到处理，最终分发会拒绝该路由。恢复后的 agent loop 仅在初始提供方/模型路由未变时保留日志中记录的推理强度；如果路由发生变化，则丢弃上一模型的不透明 ID。
当部署策略允许思考时，原生 DeepSeek 适配器声明 `off`、`high` 和 `max`，默认使用配置指定的推理强度，若未配置则使用 `high`。由适配器持有的 `off` 映射为 `thinking.type: disabled`，且不带 `reasoning_effort`；`high` 和 `max` 会启用思考并携带各自的官方协议强度值。配置为 `thinking: disabled` 的部署仅声明 `off`，并会在提供方 I/O 前拒绝启用思考的尝试。pi-ai 适配器原样发布每个确切模型的 `getSupportedThinkingLevels()` 结果，其中包括 `off`；profile 未指定默认值时保留提供方默认行为，并将提供方协议值的映射留在 pi-ai 内部。按照 pi-ai 自身 API 的要求，其通用流选项通过省略 `reasoning` 来表示 `off`。
当部署策略允许思考时，原生 DeepSeek 适配器声明 `off`、`low`、`high` 和 `max`，默认使用配置指定的推理强度，若未配置则使用 `high`。由适配器持有的 `off` 映射为 `thinking.type: disabled`，且不带 `reasoning_effort`；`low`、`high` 和 `max` 会启用思考并携带各自的同名官方协议强度值。配置为 `thinking: disabled` 的部署仅声明 `off`，并会在提供方 I/O 前拒绝启用思考的尝试。pi-ai 适配器原样发布每个确切模型的 `getSupportedThinkingLevels()` 结果，其中包括 `off`；profile 未指定默认值时保留提供方默认行为，并将提供方协议值的映射留在 pi-ai 内部。按照 pi-ai 自身 API 的要求，其通用流选项通过省略 `reasoning` 来表示 `off`。
## 备选方案