Files
marqdo/tests/ext/agent-smoke.mq.md
T
cflmy 9331af78e2 release(2.8.0): 合入 CHANGELOG/README/发版说明/skills prove 任务;CI 链接指向 doc/releases
- CHANGELOG.md:v2.8.0 条目(中英双语);
- README.md:版本行 2.7.1→2.8.0 + 愿景第 6 条(知识可证明);
- doc/releases/v2.8.0.md:双语发版说明(亮点/用法/已知限制);
- skills/marqdo-dev/index.mq.md:新增 prove 任务(路径/目录模式/
  require_lean/json;参数用 + 行声明升参,check 与 narrative-ratio 通过);
- 设计文档 §12.5 标记已合入;发版链接统一指向 doc/releases/v2.8.0.md。
2026-11-18 10:29:26 +08:00

103 lines
2.6 KiB
Markdown
Executable File

---
markup: occlusion
title: ext/agent framework smoke
description: Real llm handle from tests/ext/.env; context, extract_tool_name, run_tool via subtask.
import:
llm: ext/ai/llm.mq.md
json: lib/json.mq.md
agent: ext/ai/agent.mq.md
---
## 获取时间
**esc = > json.parse text={"v":"ok-time"}**
*> json.get value=`esc` key="v"*
# main
> llm.load_env path=.env
`工具表` =
| 工具 |
|------|
| 获取时间 |
**模型 = > llm.llm**
**助手 = > agent.agent model=`模型` tools=`工具表` standing="you are a readable Marqdo agent"**
The allowlist names what the agent may call; ToolSpec registration declares the
actual capability. This deterministic read tool takes no arguments and keeps
this smoke test independent of a real clock or mutating effect ledger.
`spec` =
| id | version | description | effect |
|----|---------|-------------|--------|
| 获取时间 | "1" | deterministic smoke value | read |
> agent.tools_clear
> agent.tools_register spec=`spec`
**id = > json.get value=`助手` key="id"**
**ctx = > agent.build_step_context agent=`助手` task="probe"**
**has_skill = > split value=`ctx` sep="Marqdo"**
**sk_n = > len value=`has_skill`**
1. `sk_n` > 1
> print text=skill-ok
2. *
> print text=skill-missing
**has_tools = > split value=`ctx` sep="获取时间"**
**tool_n = > len value=`has_tools`**
1. `tool_n` > 1
> print text=tools-ok
2. *
> print text=tools-missing
**has_src = > split value=`ctx` sep="probe"**
**src_n = > len value=`has_src`**
1. `src_n` > 1
> print text=source-ok
2. *
> print text=source-missing
**has_act = > split value=`ctx` sep="CALL:"**
**act_n = > len value=`has_act`**
1. `act_n` > 1
> print text=protocol-ok
2. *
> print text=protocol-missing
**samples = > json.parse text={"r":"CALL:获取时间","m":"think\nCALL:获取时间"}**
**raw1 = > json.get value=`samples` key="r"**
**name1 = > agent.extract_tool_name reply=`raw1`**
> print text=`name1`
**raw2 = > json.get value=`samples` key="m"**
**name2 = > agent.extract_tool_name reply=`raw2`**
> print text=`name2`
**tool_out = > agent.run_tool tools=`工具表` name="获取时间"**
> print text=`tool_out`
**turn = > json.parse text={"role":"user","content":"one"}**
> agent_history_append id=`id` item=`turn`
**turn2 = > json.parse text={"role":"assistant","content":"two"}**
> agent_history_append id=`id` item=`turn2`
**h1 = > agent_history_get id=`id`**
**n1 = > len value=`h1`**
> print text=`n1`
> `助手`.clear_history
**h2 = > agent_history_get id=`id`**
**n2 = > len value=`h2`**
> print text=`n2`
**site = > agent_call_site**
**fn = > json.get value=`site` key="function"**
> print text=`fn`