287 lines
6.8 KiB
JSON
287 lines
6.8 KiB
JSON
{
|
|
"skill": "le-wechat-article",
|
|
"date": "2026-11-05",
|
|
"threshold": 0.3,
|
|
"prediction_mode": "coverage_fallback",
|
|
"note": "词面冒烟测试;行为结论需真实对话实测(live_turns 字段手工补记)",
|
|
"cases": [
|
|
{
|
|
"text": "帮我写篇公众号文章,讲讲最近用 Agent Skill 的体会",
|
|
"family": "explicit_new_article",
|
|
"bucket": "should_trigger",
|
|
"coverage": 0.4,
|
|
"coverage_hits": [
|
|
"众号",
|
|
"公众",
|
|
"写篇",
|
|
"号文",
|
|
"文章",
|
|
"篇公"
|
|
],
|
|
"families_hit": [],
|
|
"negative_hit": false,
|
|
"predicted_trigger": true,
|
|
"expected_trigger": true,
|
|
"verdict": "pass"
|
|
},
|
|
{
|
|
"text": "把这份调研笔记整理成一篇推文",
|
|
"family": "material_to_article",
|
|
"bucket": "should_trigger",
|
|
"coverage": 0.615,
|
|
"coverage_hits": [
|
|
"一篇",
|
|
"把这",
|
|
"推文",
|
|
"整理",
|
|
"理成",
|
|
"笔记",
|
|
"调研",
|
|
"这份"
|
|
],
|
|
"families_hit": [],
|
|
"negative_hit": false,
|
|
"predicted_trigger": true,
|
|
"expected_trigger": true,
|
|
"verdict": "pass"
|
|
},
|
|
{
|
|
"text": "发公众号:ZGLM-5.3 发布了,主要更新这三点",
|
|
"family": "release_brief",
|
|
"bucket": "should_trigger",
|
|
"coverage": 0.333,
|
|
"coverage_hits": [
|
|
"众号",
|
|
"公众",
|
|
"发公",
|
|
"发布"
|
|
],
|
|
"families_hit": [],
|
|
"negative_hit": false,
|
|
"predicted_trigger": true,
|
|
"expected_trigger": true,
|
|
"verdict": "pass"
|
|
},
|
|
{
|
|
"text": "我这里有一段上次做自动化测试的复盘,写成公众号文章吧",
|
|
"family": "retro_article",
|
|
"bucket": "should_trigger",
|
|
"coverage": 0.348,
|
|
"coverage_hits": [
|
|
"一段",
|
|
"众号",
|
|
"公众",
|
|
"写成",
|
|
"号文",
|
|
"复盘",
|
|
"成公",
|
|
"文章"
|
|
],
|
|
"families_hit": [],
|
|
"negative_hit": false,
|
|
"predicted_trigger": true,
|
|
"expected_trigger": true,
|
|
"verdict": "pass"
|
|
},
|
|
{
|
|
"text": "这篇旧稿帮我改成公众号的版本,去一下 AI 味",
|
|
"family": "rewrite_old_draft",
|
|
"bucket": "should_trigger",
|
|
"coverage": 0.4,
|
|
"coverage_hits": [
|
|
"众号",
|
|
"公众",
|
|
"成公",
|
|
"旧稿",
|
|
"篇旧",
|
|
"这篇"
|
|
],
|
|
"families_hit": [],
|
|
"negative_hit": false,
|
|
"predicted_trigger": true,
|
|
"expected_trigger": true,
|
|
"verdict": "pass"
|
|
},
|
|
{
|
|
"text": "润色下这篇公众号稿子,看看能不能发",
|
|
"family": "polish_check",
|
|
"bucket": "should_trigger",
|
|
"coverage": 0.571,
|
|
"coverage_hits": [
|
|
"下这",
|
|
"众号",
|
|
"公众",
|
|
"号稿",
|
|
"润色",
|
|
"篇公",
|
|
"色下",
|
|
"这篇"
|
|
],
|
|
"families_hit": [],
|
|
"negative_hit": false,
|
|
"predicted_trigger": true,
|
|
"expected_trigger": true,
|
|
"verdict": "pass"
|
|
},
|
|
{
|
|
"text": "素材在这个目录里,整理成公众号文章,输出到 articles 下面",
|
|
"family": "dir_material",
|
|
"bucket": "should_trigger",
|
|
"coverage": 0.471,
|
|
"coverage_hits": [
|
|
"众号",
|
|
"公众",
|
|
"号文",
|
|
"成公",
|
|
"整理",
|
|
"文章",
|
|
"理成",
|
|
"素材"
|
|
],
|
|
"families_hit": [],
|
|
"negative_hit": false,
|
|
"predicted_trigger": true,
|
|
"expected_trigger": true,
|
|
"verdict": "pass"
|
|
},
|
|
{
|
|
"text": "写完把这篇同步到公众号草稿箱",
|
|
"family": "sync_draft",
|
|
"bucket": "should_trigger",
|
|
"coverage": 0.769,
|
|
"coverage_hits": [
|
|
"众号",
|
|
"公众",
|
|
"到公",
|
|
"号草",
|
|
"同步",
|
|
"把这",
|
|
"步到",
|
|
"稿箱",
|
|
"草稿",
|
|
"这篇"
|
|
],
|
|
"families_hit": [],
|
|
"negative_hit": false,
|
|
"predicted_trigger": true,
|
|
"expected_trigger": true,
|
|
"verdict": "pass"
|
|
},
|
|
{
|
|
"text": "帮我把这篇稿子群发给所有粉丝",
|
|
"family": "mass_send",
|
|
"bucket": "should_not_trigger",
|
|
"coverage": 0.25,
|
|
"coverage_hits": [
|
|
"把这",
|
|
"群发",
|
|
"这篇"
|
|
],
|
|
"families_hit": [],
|
|
"negative_hit": true,
|
|
"predicted_trigger": false,
|
|
"expected_trigger": false,
|
|
"verdict": "pass"
|
|
},
|
|
{
|
|
"text": "把这篇小红书笔记改成知乎回答",
|
|
"family": "other_platform",
|
|
"bucket": "should_not_trigger",
|
|
"coverage": 0.462,
|
|
"coverage_hits": [
|
|
"小红",
|
|
"把这",
|
|
"知乎",
|
|
"笔记",
|
|
"红书",
|
|
"这篇"
|
|
],
|
|
"families_hit": [],
|
|
"negative_hit": true,
|
|
"predicted_trigger": false,
|
|
"expected_trigger": false,
|
|
"verdict": "pass"
|
|
},
|
|
{
|
|
"text": "帮项目写一份 README 文档",
|
|
"family": "tech_doc",
|
|
"bucket": "should_not_trigger",
|
|
"coverage": 0.143,
|
|
"coverage_hits": [
|
|
"文档"
|
|
],
|
|
"families_hit": [],
|
|
"negative_hit": true,
|
|
"predicted_trigger": false,
|
|
"expected_trigger": false,
|
|
"verdict": "pass"
|
|
},
|
|
{
|
|
"text": "把这周的代码变更总结成周报",
|
|
"family": "weekly_report",
|
|
"bucket": "should_not_trigger",
|
|
"coverage": 0.417,
|
|
"coverage_hits": [
|
|
"代码",
|
|
"变更",
|
|
"周报",
|
|
"把这",
|
|
"码变"
|
|
],
|
|
"families_hit": [],
|
|
"negative_hit": true,
|
|
"predicted_trigger": false,
|
|
"expected_trigger": false,
|
|
"verdict": "pass"
|
|
},
|
|
{
|
|
"text": "给这个函数写个注释,解释一下逻辑",
|
|
"family": "code_comment",
|
|
"bucket": "should_not_trigger",
|
|
"coverage": 0.0,
|
|
"coverage_hits": [],
|
|
"families_hit": [],
|
|
"negative_hit": false,
|
|
"predicted_trigger": false,
|
|
"expected_trigger": false,
|
|
"verdict": "pass"
|
|
},
|
|
{
|
|
"text": "帮我找几个公众号选题",
|
|
"family": "topic_only",
|
|
"bucket": "near_neighbor",
|
|
"coverage": 0.375,
|
|
"coverage_hits": [
|
|
"个公",
|
|
"众号",
|
|
"公众"
|
|
],
|
|
"families_hit": [],
|
|
"negative_hit": false,
|
|
"predicted_trigger": true,
|
|
"expected_trigger": null,
|
|
"verdict": "boundary"
|
|
},
|
|
{
|
|
"text": "给这篇文章配几张图",
|
|
"family": "image_only",
|
|
"bucket": "near_neighbor",
|
|
"coverage": 0.25,
|
|
"coverage_hits": [
|
|
"文章",
|
|
"这篇"
|
|
],
|
|
"families_hit": [],
|
|
"negative_hit": false,
|
|
"predicted_trigger": false,
|
|
"expected_trigger": null,
|
|
"verdict": "boundary"
|
|
}
|
|
],
|
|
"summary": {
|
|
"scored": 13,
|
|
"mismatch": 0,
|
|
"pass_rate": 1.0
|
|
}
|
|
}
|