Update nexus: fix conflicts and sync local changes
This commit is contained in:
@@ -1,26 +1,26 @@
|
||||
---
|
||||
title: "Agent Routing Rules"
|
||||
type: concept
|
||||
last_updated: 2026-04-10
|
||||
---
|
||||
|
||||
## Definition
|
||||
Agent Routing Rules 是 OpenClaw 中绑定特定 Channel(如 Telegram)到特定模型的配置规则,优先级高于全局配置文件(`openclaw.json`)。
|
||||
|
||||
## Key Characteristics
|
||||
- 定义在 OpenClaw 的 Agent 路由层,不在 `openclaw.json` 全局 compaction 配置里
|
||||
- 修改全局配置对 Agent 级别路由无效
|
||||
- 模型 context window 直接影响可用 token 数量
|
||||
|
||||
## Common Failure Pattern
|
||||
当 fallback 机制切换到小 context 模型,且该模型在路由规则中被绑定到某个 channel 时:
|
||||
- Telegram channel → deepseek-reasoner (16K)
|
||||
- 16K context + 16K safeguard 预留 = 0 可用 token
|
||||
|
||||
## Related
|
||||
- [[Model-Fallback]]: 触发模型切换的机制
|
||||
- [[Compaction]]: Agent 级别 compaction 与全局配置的区别
|
||||
- [[Layered-Configuration]]: 分层配置的重要性
|
||||
|
||||
## Sources
|
||||
- [[养虾日记4-一次「context-limit-exceeded」错误排查-我以为是小问题-结果踩了大坑]]
|
||||
---
|
||||
title: "Agent Routing Rules"
|
||||
type: concept
|
||||
last_updated: 2026-04-10
|
||||
---
|
||||
|
||||
## Definition
|
||||
Agent Routing Rules 是 OpenClaw 中绑定特定 Channel(如 Telegram)到特定模型的配置规则,优先级高于全局配置文件(`openclaw.json`)。
|
||||
|
||||
## Key Characteristics
|
||||
- 定义在 OpenClaw 的 Agent 路由层,不在 `openclaw.json` 全局 compaction 配置里
|
||||
- 修改全局配置对 Agent 级别路由无效
|
||||
- 模型 context window 直接影响可用 token 数量
|
||||
|
||||
## Common Failure Pattern
|
||||
当 fallback 机制切换到小 context 模型,且该模型在路由规则中被绑定到某个 channel 时:
|
||||
- Telegram channel → deepseek-reasoner (16K)
|
||||
- 16K context + 16K safeguard 预留 = 0 可用 token
|
||||
|
||||
## Related
|
||||
- [[Model-Fallback]]: 触发模型切换的机制
|
||||
- [[Compaction]]: Agent 级别 compaction 与全局配置的区别
|
||||
- [[Layered-Configuration]]: 分层配置的重要性
|
||||
|
||||
## Sources
|
||||
- [[养虾日记4-一次「context-limit-exceeded」错误排查-我以为是小问题-结果踩了大坑]]
|
||||
|
||||
Reference in New Issue
Block a user