Repository navigation
Conversation
基准强度设为 xhigh 后,poteto-mode 给最难的工作再升一档,会在 Fable、 GPT-6 Astra 这类最贵的模型上用到 max。现有上限只有 xagent_effort_caps, 管不到原生 Agent 和 T3 目标模型。 新增共享字段 reasoning_effort_cap(如 "xhigh"):任何宿主上任何 worker 在按任务调档之后都不超过它;只降强度,不换模型,按难度选模型照旧。 只能写在共享层,null 删除。T3 解析器在代码里统一降档并给出警告,先套 全局上限、再套 xagent 自己的上限;原生主控按 poteto-mode 的强度规则 执行。runtime、t3/antigravity 适配、setup、中英文 README、ADAPTATION 同步,fork 守卫锚定 poteto-mode 的升档句。新测试在 main 的 bin/ 上失败。 Co-Authored-By: Claude Opus 5.5 <[email protected]>
Co-Authored-By: Claude Opus 5.5 <[email protected]>
全局 reasoning_effort_cap 会把所有模型都封顶,连普通角色上"最微妙的
工作"在 Opus/Sol 上升到 max 也一起卡掉;实际只想限制最贵的 Fable 和
Astra。改为共享字段 model_effort_caps(如 {"fable": "xhigh", "astra":
"xhigh"}),键是模型名里以短横线分隔的一段,匹配 fable 别名、
claude-fable-5-1、gpt-6-astra 等所有写法;其他模型照常升档。T3 解析器
对解析出的模型或 xagent 固定的模型先套模型上限、再套 xagent 上限;原生
主控按 poteto-mode 强度规则执行。新测试在 main 的 bin/ 上失败。
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Owner
Author
|
不合并:最难任务只在被判为最难的那一步子任务上用 Fable/Astra(主线程不切换,子任务也不再往下派任务),用户决定保留升到 max 的能力,不需要强度上限。 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
问题
共享基准强度已统一为
xhigh,原生和 T3 都按难度切换模型:最难的工作用 Fable 5.1 或 GPT-6 Astra。poteto-mode 的强度规则会给“最难或最微妙的工作”在基准上再升一档,于是这两个最贵的模型会跑到max。现有的xagent_effort_caps只管外部 xagent,原生 Agent 工具和 T3 目标模型都没有上限可设。目标是只卡 Fable 和 Astra。其他模型(例如普通角色上的 Opus/Sol 处理很微妙的工作时)仍可以升到
max。改动
model_effort_caps(如{"fable": "xhigh", "astra": "xhigh"})。键是模型名里以短横线分隔的一段,能匹配任何宿主上该模型的各种写法:fable别名、claude-fable-5-1、gpt-6-astra、group/auto-gpt-6-astra。只降强度,不换模型,按难度选模型照旧。只能写在共享层,宿主覆盖会被拒绝,写null即删除。不设时行为与上游一致。max或该模型的上限为止。reasoning_effort_cap,会把所有模型都封顶,b0eb9c5 已替换为按模型的上限。effort-cap-raise,防止上游同步把升档句改回“升到max”。0.15.25-shared.10。验证
bash scripts/check.sh runtime→runtime checks passed.(Node 88 条,fork 守卫 26 条)cd3c5d0)的bin/跑同一份测试:61 pass / 2 fail,恰好是 2 个新用例{"fable":"xhigh","astra":"xhigh"}后:hardest tasks --effort max→ Fable 5.1 @ xhigh;最难的架构组 → Fable @ xhigh、Astra @ xhigh、gemini @ high;bug-fix --effort max→ Opus 5.5 @ max(不受影响);feature基准 → Opus @ xhigh限制
xagent:gemini),无法按模型匹配,只受xagent_effort_caps限制。model_effort_caps写进~/.config/pstack/config.json才生效。🤖 Generated with Claude Code