senpi-trading-runtime

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Senpi Trading Runtime — OpenClaw Plugin

On-chain position tracker with automated DSL (Dynamic Stop-Loss) exit engine. Monitors a wallet's positions on Hyperliquid for lifecycle events (open, close, edit, flip) and applies two-phase trailing stop-loss protection to all positions. It links to an existing strategy wallet address and does not create wallets.

链上仓位跟踪器，搭载自动化DSL（动态止损）退出引擎。监控Hyperliquid上钱包的仓位生命周期事件（开仓、平仓、编辑、方向翻转），为所有仓位应用两阶段移动止损保护。它会关联已有的策略钱包地址，不会自行创建钱包。

Core Concepts

核心概念

Flow: Position Tracker scanner → detects position changes → DSL exit engine manages trailing stops

Position Tracker (
```
position_tracker
```
scanner) polls the wallet on-chain, detecting opens, closes, increases, decreases, and flips.
DSL exit engine monitors tracked positions on a timer, computing trailing stop-loss floors across two phases.
Strategy (in runtime) = link to an existing wallet address. The wallet address is the strategy identifier everywhere, but wallet creation happens via Senpi MCP (not the runtime).

Key insight: The position tracker enables DSL protection for ALL positions of a strategy address — including those opened manually on the exchange or by other tools.

流程： Position Tracker扫描器 → 检测仓位变化 → DSL退出引擎管理移动止损

Position Tracker（
```
position_tracker
```
扫描器）轮询链上钱包，检测开仓、平仓、仓位增加、仓位减少和方向翻转。
DSL退出引擎定时监控被跟踪的仓位，跨两个阶段计算移动止损底线。
运行时内的策略 = 关联已有钱包地址。钱包地址是全平台通用的策略标识，但钱包创建需要通过Senpi MCP完成，而非运行时本身。

核心要点： 仓位跟踪器可为策略地址的所有仓位提供DSL保护——包括在交易所手动开仓或通过其他工具开仓的仓位。

Wallet lifecycle (required order)

钱包生命周期（必须遵循顺序）

Create strategy wallet address via Senpi MCP calls (or confirm it already exists).
Put that wallet address into runtime YAML (
```
strategies.<name>.wallet
```
/
```
${WALLET_ADDRESS}
```
).
Install/create runtime. This links the runtime to the existing wallet for monitoring and exits.

Never treat

openclaw senpi runtime create

as wallet creation.

通过Senpi MCP调用创建策略钱包地址（或确认该地址已存在）。
将该钱包地址填入运行时YAML的
```
strategies.<name>.wallet
```
/
```
${WALLET_ADDRESS}
```
字段。
安装/创建运行时。这一步会将运行时与已有钱包关联，用于监控和退出操作。

绝对不要将

openclaw senpi runtime create

命令当作钱包创建操作。

Strategy wallet validation (mandatory before runtime create/install)

策略钱包校验（创建/安装运行前必须执行）

Use this exact MCP-first guardrail to avoid using embedded wallets by mistake:

Call

strategy_list

and collect

strategies[].strategyWalletAddress

A runtime wallet is valid only if it appears in that list (case-insensitive match).
If list is empty OR provided wallet is not in list: call
```
strategy_create_custom_strategy
```
to create a strategy wallet first.
Use the newly created strategy wallet address directly as
```
WALLET_ADDRESS
```
in runtime YAML, then install runtime.

Hard rules:

Never use embedded wallet/injected wallet for runtime linking.
Never treat a wallet as a strategy wallet unless it exists in
```
strategy_list
```
.
Runtime install is blocked until strategy wallet validation passes.
Always confirm with the user before creating a new strategy wallet address, and explicitly confirm the budget (
```
initialBudget
```
) that will be used.

Example MCP flow (tool names from Senpi MCP):

text

strategy_list({})
if provided_wallet not in strategies[].strategyWalletAddress:
  # confirm with user first: create new strategy wallet + initialBudget
  strategy_create_custom_strategy({
    initialBudget: <budget_usdc>,
    positions: [],
    skillName: <strategy_or_runtime_name>,
    skillVersion: "1.0.0"
  })

Notes:

```
initialBudget
```
is required when creating the strategy wallet.
Use
```
positions: []
```
when you only need the strategy wallet created now and will trade later.

请严格遵循以下MCP优先的校验规则，避免误绑定内嵌钱包：

调用

strategy_list

接口，收集

strategies[].strategyWalletAddress

列表。

只有当待绑定的钱包出现在上述列表中时才视为有效（不区分大小写匹配）。
如果列表为空，或者待绑定钱包不在列表中：请先调用
```
strategy_create_custom_strategy
```
创建策略钱包。
将新创建的策略钱包地址直接作为运行时YAML中的
```
WALLET_ADDRESS
```
使用，再安装运行时。

硬性规则：

绝对不要使用内嵌钱包/注入钱包进行运行时绑定。
除非钱包存在于
```
strategy_list
```
返回结果中，否则不要将其视为策略钱包。
策略钱包校验通过前，禁止安装运行时。
创建新策略钱包地址前必须与用户确认，且明确确认将使用的预算（
```
initialBudget
```
）。

MCP流程示例（工具名来自Senpi MCP）：

text

strategy_list({})
if provided_wallet not in strategies[].strategyWalletAddress:
  # 先和用户确认：创建新策略钱包 + 初始预算
  strategy_create_custom_strategy({
    initialBudget: <budget_usdc>,
    positions: [],
    skillName: <strategy_or_runtime_name>,
    skillVersion: "1.0.0"
  })

注意事项：

创建策略钱包时
```
initialBudget
```
为必填项。
如果你只需要创建策略钱包后续再交易，可传入
```
positions: []
```
。

CLI Commands

CLI命令

All commands require the OpenClaw gateway running (

openclaw gateway run

Use

openclaw senpi --cheatsheet

to print the full plugin command cheatsheet to stdout.

所有命令都需要先启动OpenClaw网关（

openclaw gateway run

）。

使用

openclaw senpi --cheatsheet

可在标准输出打印完整的插件命令速查表。

Runtime management

运行时管理

Prerequisite: strategy wallet address already exists (created/fetched via Senpi MCP).

bash

undefined

前置条件：策略钱包地址已存在（通过Senpi MCP创建/获取）。

bash

undefined

Create a runtime from YAML file

从YAML文件创建运行时

openclaw senpi runtime create --path ./my-strategy.yaml openclaw senpi runtime create -p ./my-strategy.yaml # short form

openclaw senpi runtime create --path ./my-strategy.yaml openclaw senpi runtime create -p ./my-strategy.yaml # 简写形式

Create with inline YAML content

用内联YAML内容创建

openclaw senpi runtime create --content "<yaml>" openclaw senpi runtime create -c "<yaml>" # short form

openclaw senpi runtime create --content "<yaml>" openclaw senpi runtime create -c "<yaml>" # 简写形式

Create with custom ID

创建时指定自定义ID

openclaw senpi runtime create --path ./my-strategy.yaml --runtime-id my-name

List installed runtimes (id, wallet, source, status)

列出已安装的运行时（ID、钱包、来源、状态）

openclaw senpi runtime list

Delete a runtime

删除运行时

openclaw senpi runtime delete <runtime_id> # positional id openclaw senpi runtime delete --id <runtime_id> # named flag

undefined

openclaw senpi runtime delete <runtime_id> # 位置参数传ID openclaw senpi runtime delete --id <runtime_id> # 命名参数传ID

undefined

DSL position inspection

DSL仓位查看

bash

undefined

bash

undefined

All active DSL-tracked positions

所有被DSL跟踪的活跃仓位

openclaw senpi dsl positions openclaw senpi dsl positions -r <id> # filter by runtime id openclaw senpi dsl positions -a <addr> # filter by wallet address openclaw senpi dsl positions --json

openclaw senpi dsl positions openclaw senpi dsl positions -r <id> # 按运行时ID过滤 openclaw senpi dsl positions -a <addr> # 按钱包地址过滤 openclaw senpi dsl positions --json

Inspect one position (full DslState)

查看单个仓位的完整DslState

openclaw senpi dsl inspect <ASSET> openclaw senpi dsl inspect SOL -r <id> openclaw senpi dsl inspect SOL -a <addr> openclaw senpi dsl inspect BTC --json

Archived (closed) positions — reason, ROE, phase, tier

已归档（已平仓）仓位：平仓原因、ROE、阶段、层级

openclaw senpi dsl closes openclaw senpi dsl closes -r <id> openclaw senpi dsl closes -a <addr> openclaw senpi dsl closes -l <n> openclaw senpi dsl closes --json

undefined

openclaw senpi dsl closes openclaw senpi dsl closes -r <id> openclaw senpi dsl closes -a <addr> openclaw senpi dsl closes -l <n> openclaw senpi dsl closes --json

undefined

Runtime diagnostics

运行时诊断

bash

openclaw senpi status                    # lightweight health summary (all runtimes)
openclaw senpi status -r <id>
openclaw senpi status --json

openclaw senpi state                     # full operational state (all runtimes)
openclaw senpi state -r <id>
openclaw senpi state --json

bash

openclaw senpi status                    # 轻量级健康摘要（所有运行时）
openclaw senpi status -r <id>
openclaw senpi status --json

openclaw senpi state                     # 完整运行状态（所有运行时）
openclaw senpi state -r <id>
openclaw senpi state --json

In-shell reference (

senpi guide

)

内置参考（

senpi guide

）

bash

openclaw senpi guide                 # overview + quick command list
openclaw senpi guide scanners        # scanner types and config fields
openclaw senpi guide actions         # action types and decision modes
openclaw senpi guide dsl             # DSL exit engine: phases, tiers, time cuts
openclaw senpi guide examples        # print minimal strategy YAML to stdout
openclaw senpi guide schema          # full YAML schema field reference
openclaw senpi guide version         # plugin version and changelog URL

bash

openclaw senpi guide                 # 概览 + 快速命令列表
openclaw senpi guide scanners        # 扫描器类型和配置字段
openclaw senpi guide actions         # 动作类型和决策模式
openclaw senpi guide dsl             # DSL退出引擎：阶段、层级、定时减仓
openclaw senpi guide examples        # 在标准输出打印最简策略YAML示例
openclaw senpi guide schema          # 完整YAML Schema字段参考
openclaw senpi guide version         # 插件版本和更新日志URL

CLI preferences (

senpi config

)

CLI偏好设置（

senpi config

）

Persists to

~/.openclaw/senpi-cli.json

. Restart gateway to apply.

bash

openclaw senpi config set-chat-id <chatId>
openclaw senpi config set-senpi-jwt-token <token>
openclaw senpi config set-state-dir <dir>
openclaw senpi config get <key>      # telegram-chat-id | senpi-jwt-token | state-dir
openclaw senpi config list           # secrets masked
openclaw senpi config unset <key>
openclaw senpi config reset

配置持久化存储在

~/.openclaw/senpi-cli.json

，重启网关后生效。

bash

openclaw senpi config set-chat-id <chatId>
openclaw senpi config set-senpi-jwt-token <token>
openclaw senpi config set-state-dir <dir>
openclaw senpi config get <key>      # 可查询字段：telegram-chat-id | senpi-jwt-token | state-dir
openclaw senpi config list           # 密钥会被脱敏
openclaw senpi config unset <key>
openclaw senpi config reset

Gateway RPC (advanced)

网关RPC（高级功能）

Common calls:

bash

openclaw gateway call senpi.installRuntime --params '{"runtimeYamlContent":"..."}'
openclaw gateway call senpi.listRuntimes --json --params '{}'
openclaw gateway call senpi.deleteRuntime --params '{"id":"my-runtime"}'
openclaw gateway call senpi.listDslPositions --json --params '{}'
openclaw gateway call senpi.getDslPositionState --json --params '{"asset":"SOL"}'
openclaw gateway call senpi.listDslArchives --json --params '{"limit":20}'
openclaw gateway call senpi.getHealthStatus --json --params '{}'
openclaw gateway call senpi.getSystemState --json --params '{}'

Methods summary

Method	Params	Success response
`senpi.installRuntime`	`runtimeYamlContent` or `runtimeYamlPath`	Runtime installed
`senpi.listRuntimes`	—	List of runtimes
`senpi.deleteRuntime`	`id` or address	Deleted
`senpi.listDslPositions`	`runtimeId?`	Active positions
`senpi.getDslPositionState`	`asset` , `runtimeId?`	Single position state
`senpi.listDslArchives`	`runtimeId?` , `address?` , `limit?`	`closes[]` (asset, direction, entryPrice, lastPrice, currentROE, closeReason, closedAt, phase, …)
`senpi.getHealthStatus`	`runtimeId?` (string)	`status` or `statuses[]` (health, components.scanners, components.dsl)
`senpi.getSystemState`	`runtimeId?` (string)	`state` or `states[]` (health, stateDir, scanner/DSL components)

Close reasons for

listDslArchives

dsl_breach

hard_timeout

weak_peak_cut

dead_weight_cut

exchange_sl_hit

manual

常用调用：

bash

openclaw gateway call senpi.installRuntime --params '{"runtimeYamlContent":"..."}'
openclaw gateway call senpi.listRuntimes --json --params '{}'
openclaw gateway call senpi.deleteRuntime --params '{"id":"my-runtime"}'
openclaw gateway call senpi.listDslPositions --json --params '{}'
openclaw gateway call senpi.getDslPositionState --json --params '{"asset":"SOL"}'
openclaw gateway call senpi.listDslArchives --json --params '{"limit":20}'
openclaw gateway call senpi.getHealthStatus --json --params '{}'
openclaw gateway call senpi.getSystemState --json --params '{}'

方法汇总

方法	参数	成功响应
`senpi.installRuntime`	`runtimeYamlContent` 或 `runtimeYamlPath`	运行时已安装
`senpi.listRuntimes`	—	运行时列表
`senpi.deleteRuntime`	`id` 或地址	已删除
`senpi.listDslPositions`	`runtimeId?`	活跃仓位列表
`senpi.getDslPositionState`	`asset` , `runtimeId?`	单个仓位状态
`senpi.listDslArchives`	`runtimeId?` , `address?` , `limit?`	`closes[]` （包含资产、方向、入场价、最新价、当前ROE、平仓原因、平仓时间、阶段等字段）
`senpi.getHealthStatus`	`runtimeId?` （字符串）	`status` 或 `statuses[]` （健康状态、scanners组件状态、dsl组件状态）
`senpi.getSystemState`	`runtimeId?` （字符串）	`state` 或 `states[]` （健康状态、状态目录、扫描器/DSL组件状态）

listDslArchives

的平仓原因枚举：

dsl_breach

（触发DSL止损）、

hard_timeout

（硬超时）、

weak_peak_cut

（弱峰值减仓）、

dead_weight_cut

（无效仓位减仓）、

exchange_sl_hit

（触发交易所止损）、

manual

（手动平仓）。

Runtime YAML

运行时YAML

The runtime YAML drives all behavior. Top-level keys:

name

strategies

scanners

actions

exit

notifications

yaml

name: my-tracker
version: 1.0.0
description: >
  On-chain position tracker with DSL trailing stop-loss.

strategies:
  main:
    wallet: "${WALLET_ADDRESS}"
    budget: 500
    slots: 2
    margin_per_slot: 200
    trading_risk: conservative    # conservative | moderate | aggressive
    enabled: true

scanners:
  - name: position_tracker
    type: position_tracker
    interval: 10s

actions:
  - name: position_tracker_action
    action_type: POSITION_TRACKER
    decision_mode: rule
    scanners: [position_tracker]

exit:
  engine: dsl
  interval_seconds: 30            # how often the price monitor runs (5-3600)
  dsl_preset:                     # single preset (no named map needed)
    hard_timeout:
      enabled: true
      interval_in_minutes: 360
    weak_peak_cut:
      enabled: true
      interval_in_minutes: 120
      min_value: 5
    dead_weight_cut:
      enabled: true
      interval_in_minutes: 60
    phase1:
      enabled: true
      max_loss_pct: 4.0
      retrace_threshold: 7
      consecutive_breaches_required: 1
    phase2:
      enabled: true
      tiers:
        - { trigger_pct: 7,  lock_hw_pct: 40 }
        - { trigger_pct: 12, lock_hw_pct: 55 }
        - { trigger_pct: 15, lock_hw_pct: 75 }
        - { trigger_pct: 20, lock_hw_pct: 85 }

notifications:
  telegram_chat_id: "${TELEGRAM_CHAT_ID}"

Environment variable substitution:

${VAR}

and

${VAR:-default}

resolved at load time.

For full field details: YAML Schema Reference

运行时YAML定义所有运行逻辑。顶层字段包括：

name

、

strategies

、

scanners

、

actions

、

exit

、

notifications

。

yaml

name: my-tracker
version: 1.0.0
description: >
  On-chain position tracker with DSL trailing stop-loss.

strategies:
  main:
    wallet: "${WALLET_ADDRESS}"
    budget: 500
    slots: 2
    margin_per_slot: 200
    trading_risk: conservative    # conservative | moderate | aggressive
    enabled: true

scanners:
  - name: position_tracker
    type: position_tracker
    interval: 10s

actions:
  - name: position_tracker_action
    action_type: POSITION_TRACKER
    decision_mode: rule
    scanners: [position_tracker]

exit:
  engine: dsl
  interval_seconds: 30            # 价格监控运行频率（单位秒，范围5-3600）
  dsl_preset:                     # 单个预设，无需命名映射
    hard_timeout:
      enabled: true
      interval_in_minutes: 360
    weak_peak_cut:
      enabled: true
      interval_in_minutes: 120
      min_value: 5
    dead_weight_cut:
      enabled: true
      interval_in_minutes: 60
    phase1:
      enabled: true
      max_loss_pct: 4.0
      retrace_threshold: 7
      consecutive_breaches_required: 1
    phase2:
      enabled: true
      tiers:
        - { trigger_pct: 7,  lock_hw_pct: 40 }
        - { trigger_pct: 12, lock_hw_pct: 55 }
        - { trigger_pct: 15, lock_hw_pct: 75 }
        - { trigger_pct: 20, lock_hw_pct: 85 }

notifications:
  telegram_chat_id: "${TELEGRAM_CHAT_ID}"

环境变量替换：加载时会解析

${VAR}

和

${VAR:-default}

格式的变量。

完整字段说明请参考：YAML Schema Reference

DSL Exit Engine — Key Concepts

DSL退出引擎——核心概念

Two-phase trailing stop-loss protecting open positions:

Phase 1 (from entry until the first tier is reached):

Absolute floor from
```
max_loss_pct
```
(cap on loss from entry, scaled by leverage)
Trailing floor from
```
retrace_threshold
```
(ROE % pullback from high-water mark)
Effective floor = stricter of both (LONG: higher price; SHORT: lower)
Optional time-based cuts at preset level:
```
hard_timeout
```
,
```
weak_peak_cut
```
,
```
dead_weight_cut
```
Exchange SL stays at the absolute floor only; tighter exits enforced by runtime

Phase 2 (after the first tier is reached — Phase 2 always starts at tier 0):

Tier floor =
```
lock_hw_pct
```
% of current high-water ROE (trails as HW advances, never loosens)
Exchange SL tracks the full effective floor and updates as it moves

Tiers are profit milestones (by ROE % from entry). Each tier has two fields:

trigger_pct

(ROE % that activates it) and

lock_hw_pct

(% of high-water ROE to lock as floor). Tiers must have strictly increasing

trigger_pct

Consecutive breaches are tick-based (not time-based). Each monitor tick (every

interval_seconds

), if price violates the floor, the breach counter increments. Recovery resets it.

Retrace convention:

retrace_threshold

is ROE %. Engine converts to price:

retrace_ROE% / 100 / leverage

. At 10x leverage, 7% ROE retrace = 0.7% price below high-water.

For full DSL configuration with all fields, time-based cuts, close reasons, and events: DSL Configuration Reference

保护开仓仓位的两阶段移动止损机制：

阶段1（从入场到触发第一个层级前）：

由
```
max_loss_pct
```
定义绝对止损底线（入场价的最大亏损比例，按杠杆缩放）
由
```
retrace_threshold
```
定义移动止损底线（从最高水位回撤的ROE百分比）
实际生效底线为两者中更严格的那个（多头取更高价格，空头取更低价格）
可选预设触发的定时减仓规则：
```
hard_timeout
```
、
```
weak_peak_cut
```
、
```
dead_weight_cut
```
交易所侧止损仅设置为绝对底线，更严格的退出由运行时强制执行

阶段2（触发第一个层级后，阶段2始终从0级开始）：

层级底线 = 当前最高水位ROE的
```
lock_hw_pct
```
%（随最高水位上升而上移，永远不会放宽）
交易所侧止损会跟踪完整的生效底线，随底线变动同步更新

层级是盈利里程碑（按入场以来的ROE百分比计算）。每个层级包含两个字段：

trigger_pct

（激活该层级的ROE百分比）和

lock_hw_pct

（作为止损底线锁定的最高水位ROE比例）。层级的

trigger_pct

必须严格递增。

连续触发次数按tick计数（而非时间）。每个监控tick（每

interval_seconds

执行一次）如果价格跌破底线，触发计数器就会加1，价格恢复后计数器重置。

回撤计算规则：

retrace_threshold

为ROE百分比。引擎会转换为价格：

回撤ROE% / 100 / 杠杆

。10倍杠杆下，7%的ROE回撤对应价格比最高水位低0.7%。

包含所有字段、定时减仓、平仓原因和事件的完整DSL配置说明请参考：DSL Configuration Reference

Environment Variables

环境变量

Variable	Required	Purpose
`WALLET_ADDRESS`	Yes	Existing strategy wallet address from Senpi MCP (used in YAML via `${WALLET_ADDRESS}` ).
`SENPI_API_KEY`	For live MCP	Senpi MCP authentication.
`TELEGRAM_CHAT_ID`	No	Telegram chat ID for notifications.
`DSL_STATE_DIR`	No	Override DSL state file directory.

变量名	必填	用途
`WALLET_ADDRESS`	是	从Senpi MCP获取的已有策略钱包地址（在YAML中通过 `${WALLET_ADDRESS}` 引用）。
`SENPI_API_KEY`	对接正式MCP时需要	Senpi MCP鉴权凭证。
`TELEGRAM_CHAT_ID`	否	接收通知的Telegram聊天ID。
`DSL_STATE_DIR`	否	覆盖DSL状态文件的存储目录。

References

参考文档

Runtime Concepts — Conceptual explanation of scanners, actions, DSL phases, and what every field controls in trading terms
YAML Schema Reference — All runtime YAML fields and DSL preset options
DSL Configuration Reference — Full DSL exit engine: phases, tiers, time cuts, close reasons, events
Strategy Examples — Ready-to-use YAML examples with different DSL tuning profiles
Migration from DSL Cron — Upgrade from old
```
dsl-v5.py
```
cron to plugin runtime
Runtime Template — Starter runtime.yaml with field mapping comments

Runtime Concepts — 扫描器、动作、DSL阶段的概念说明，以及所有字段在交易层面的控制逻辑
YAML Schema Reference — 所有运行时YAML字段和DSL预设选项说明
DSL Configuration Reference — 完整DSL退出引擎说明：阶段、层级、定时减仓、平仓原因、事件
Strategy Examples — 可直接使用的YAML示例，包含不同的DSL调优配置
Migration from DSL Cron — 从旧版
```
dsl-v5.py
```
定时任务升级到插件运行时的迁移指南
Runtime Template — 入门级runtime.yaml模板，附带字段映射注释

senpi-trading-runtime

Original

Translation

Senpi Trading Runtime — OpenClaw Plugin

Senpi Trading Runtime — OpenClaw Plugin

Core Concepts

核心概念

Wallet lifecycle (required order)

钱包生命周期（必须遵循顺序）

Strategy wallet validation (mandatory before runtime create/install)

策略钱包校验（创建/安装运行前必须执行）

CLI Commands

CLI命令

Runtime management

运行时管理

Create a runtime from YAML file

从YAML文件创建运行时

Create with inline YAML content

用内联YAML内容创建

Create with custom ID

创建时指定自定义ID

List installed runtimes (id, wallet, source, status)

列出已安装的运行时（ID、钱包、来源、状态）

Delete a runtime

删除运行时

DSL position inspection

DSL仓位查看

All active DSL-tracked positions

所有被DSL跟踪的活跃仓位

Inspect one position (full DslState)

查看单个仓位的完整DslState

Archived (closed) positions — reason, ROE, phase, tier

已归档（已平仓）仓位：平仓原因、ROE、阶段、层级

Runtime diagnostics

运行时诊断

In-shell reference (senpi guide)

内置参考（senpi guide）

CLI preferences (senpi config)

CLI偏好设置（senpi config）

Gateway RPC (advanced)

网关RPC（高级功能）

Runtime YAML

运行时YAML

DSL Exit Engine — Key Concepts

DSL退出引擎——核心概念

Environment Variables

环境变量

References

参考文档

In-shell reference (
`senpi guide`
)

内置参考（
`senpi guide`
）

CLI preferences (
`senpi config`
)

CLI偏好设置（
`senpi config`
）