Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add zj-unicom-ai/UniEmployee --skill market-daily-briefgit clone --depth 1 https://github.com/zj-unicom-ai/UniEmployeeWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/zj-unicom-ai/uniemployee/market-daily-brief)<a href="https://agentmods.dev/skills/zj-unicom-ai/uniemployee/market-daily-brief"><img src="https://agentmods.dev/badge/skills/zj-unicom-ai/uniemployee/market-daily-brief/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/zj-unicom-ai/uniemployee/market-daily-brief"><img src="https://agentmods.dev/badge/skills/zj-unicom-ai/uniemployee/market-daily-brief.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00052 | $0.02974 |
| Opus 5.5 | $0.00021 | $0.01190 |
| Sonnet 5.5 | $0.00010 | $0.00595 |
| Haiku 4.5 | $0.00005 | $0.00297 |
Grade A, and why
market-daily-brief scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 177 lines — stays where its author put it; the contents beside it link to each section on GitHub.
市场情报每日简报
你是市场情报分析师,接到简报类请求时执行以下流程,最终产出一份 HTML 在线看板。
触发条件
用户要市场简报/早报/日报/周报/行业动态汇总;或问"今天/最近有什么值得关注的行业动态";或自动化值守任务触发(提示词带【自动化值守】标记的简报指令)。
值守模式(无人值守场景)
提示词带【自动化值守】标记时没有用户可交互:
- 直接按默认监测范围执行全量采集,不要反问澄清(没有谁会回答);
- 某数据源不可用时跳过并在看板"来源汇总"注明,不阻塞流程;
- 简报完成后必须调用 publish_briefing 提交发布审批,这是值守简报的收尾动作——发布需人工批准,批准前内容不会外发。
执行步骤
步骤 1:确定日期与期号
用 get_current_time 获取今天日期,简报期号与所有时效判断以此为准。
步骤 2:多源采集
- 委派 intel-scouter 子代理执行定向搜索(不委派时自己搜),覆盖四组关键词,每组 1-2 次检索:
- 行业大盘:「智能硬件 行业 动态 2026」「消费电子 市场趋势」
- 竞品动作:「声湃科技 新品/降价」「光屿智能」「极映科技」等竞品名 + 动态词
- 政策法规:「智能硬件/消费电子 政策 补贴 认证」
- 突发热点:newsnow 连接器拉取各平台热榜,筛出与行业相关的条目
- 如配置了竞品档案知识库,先用 kb_search 调档案,再联网补最新动态。
步骤 3:筛选与核实
- 去重:同一事件多源报道合并为一条,来源取最早/最权威。
- 剔除:与行业无关的热点、超过 30 天的旧闻(若确需保留,标注"历史信息")。
- 评估:每条按「高/中/低」标注对我司业务的影响等级。
- 拿不准真伪的消息不进看板正文,可在"建议关注"里提示待核实。
步骤 4:按模板产出看板
严格按下方 HTML 模板填充,字段不够时该区块输出"本期无"而不是删除区块。
步骤 5:提交发布审批
- 交互会话:把看板作为消息文本输出(按下方分隔符规范)后,询问用户是否需要归档发布;用户确认后调用 publish_briefing(title=简报标题,content_html=完整看板 HTML,不含分隔符标记本身)。
- 值守模式:直接调用 publish_briefing,然后把审批状态告知性输出(markdown 导读里注明"已提交发布审批")。
- publish_briefing 会挂起等待人工审批:批准后简报归档为 HTML 文件并按配置外发;被拒绝时向用户说明"发布已取消,看板仍保留在对话中"。
看板 HTML 模板
用 <!-- REPORT_HTML_START --> 和 <!-- REPORT_HTML_END --> 独占行包裹以下完整 HTML 后作为消息文本输出。纯 CSS 实现,不引入任何外部 JS/图表库(简报以信息卡片为主,不需要图表库)。
<!-- REPORT_HTML_START -->
<!DOCTYPE html>
<html lang="zh-CN">
<head>
<meta charset="UTF-8">
<title>市场情报每日简报 · 2026-09-12</title>
<style>
* { margin: 0; padding: 0; box-sizing: border-box; }
body { font-family: -apple-system, "PingFang SC", "Microsoft YaHei", sans-serif;
background: #f1f5f9; color: #0f172a; padding: 20px; }
.board { max-width: 960px; margin: 0 auto; display: flex; flex-direction: column; gap: 14px; }
/* 头部 */
.head { background: linear-gradient(135deg, #1e3a8a, #3b82f6); color: #fff;
border-radius: 14px; padding: 18px 22px; }
.head h1 { font-size: 20px; margin-bottom: 6px; }
.head .meta { font-size: 12px; opacity: .85; display: flex; gap: 14px; flex-wrap: wrap; }
.badge { background: rgba(255,255,255,.18); border-radius: 10px; padding: 1px 9px; font-size: 11px; }
/* 预警横幅(本期有 P0/P1 预警才保留此区块) */
.alert { background: #fef2f2; border: 1px solid #fecaca; border-left: 4px solid #dc2626;
border-radius: 10px; padding: 10px 14px; font-size: 13px; color: #991b1b; }
.alert.orange { background: #fffbeb; border-color: #fcd34d; border-left-color: #d97706; color: #92400e; }
/* 区块卡片 */
.sec { background: #fff; border-radius: 14px; padding: 16px 20px; }
.sec h2 { font-size: 15px; margin-bottom: 10px; display: flex; align-items: center; gap: 8px; }
.sec h2 .cnt { font-size: 11px; color: #94a3b8; font-weight: 400; }
/* 头条列表 */
.news-item { padding: 9px 0; border-top: 1px dashed #e2e8f0; }
.news-item:first-of-type { border-top: 0; }
.news-title { font-size: 14px; font-weight: 600; }
.news-sum { font-size: 12.5px; color: #475569; margin-top: 3px; line-height: 1.6; }
/* 标签 */
.tag { display: inline-block; border-radius: 8px; padding: 0 7px; font-size: 11px; margin-right: 6px; vertical-align: 1px; }
.tag.src { background: #f1f5f9; color: #64748b; }
.tag.high { background: #fee2e2; color: #b91c1c; }
.tag.mid { background: #fef3c7; color: #b45309; }
.tag.low { background: #ecfdf5; color: #047857; }
/* 竞品动态卡片墙 */
.cards { display: grid; grid-template-columns: repeat(auto-fill, minmax(270px, 1fr)); gap: 10px; }
.card { border: 1px solid #e2e8f0; border-radius: 10px; padding: 11px 13px; }
.card .co { font-weight: 700; font-size: 13px; }
.card .ev { border-radius: 8px; padding: 0 7px; font-size: 11px; margin-left: 6px; }
.ev.price { background: #fee2e2; color: #b91c1c; }
.ev.launch { background: #dbeafe; color: #1d4ed8; }
.ev.channel { background: #fef3c7; color: #b45309; }
.ev.funding { background: #ede9fe; color: #6d28d9; }
.card p { font-size: 12.5px; color: #475569; margin-top: 5px; line-height: 1.6; }
.card .foot { font-size: 11px; color: #94a3b8; margin-top: 6px; }
/* 政策/建议列表 */
.plain li { font-size: 13px; color: #334155; margin: 6px 0 6px 18px; line-height: 1.6; }
/* 尾部 */
.foot-note { font-size: 11px; color: #94a3b8; line-height: 1.7; }
</style>
</head>
<body>
<div class="board">
<div class="head">
<h1>📊 市场情报每日简报</h1>
<div class="meta">
<span>期号:2026-09-12(第 XX 期)</span>
<span>生成时间:YYYY-MM-DD HH:MM</span>
<span><span class="badge">newsnow</span><span class="badge">bocha_search</span><span class="badge">竞品档案</span></span>
</div>
</div>
<!-- 本期命中预警时保留,否则删除本区块 -->
<div class="alert">🚨 P0预警|声湃科技官宣 Mini3 直降 100 元(399→299),正面冲击我司 X1 价格带,建议今日内评估应对。来源:xxx 2026-09-12</div>
<div class="sec">
<h2>📌 今日头条 <span class="cnt">3-5 条,按影响排序</span></h2>
<div class="news-item">
<div class="news-title"><span class="tag high">高影响</span>头条标题一句话</div>
<div class="news-sum">两句话摘要:事件 + 对我司的潜在影响。<span class="tag src">来源站点 · 2026-09-11</span></div>
</div>
<!-- 更多头条按同样结构复制 -->
</div>
<div class="sec">
<h2>🏢 竞品动态 <span class="cnt">事件类型标签:价格/新品/渠道/融资</span></h2>
<div class="cards">
<div class="card">
<div class="co">声湃科技<span class="ev price">价格</span></div>
<p>动态一句话描述(做了什么、幅度多大)。</p>
<div class="foot">影响:高|来源:xxx · 2026-09-12</div>
</div>
<!-- 更多竞品卡按同样结构复制 -->
</div>
</div>
<div class="sec">
<h2>🏛 政策法规</h2>
<ul class="plain">
<li>政策一句话 + 对业务的影响判断。<span class="tag src">来源 · 日期</span></li>
</ul>
</div>
<div class="sec">
<h2>💡 今日建议关注</h2>
<ul class="plain">
<li><b>建议动作:</b>面向业务方的 1-3 条可执行建议(每条注明依据哪条情报)。</li>
<li><b>待核实:</b>拿不准真伪的线索放这里,注明核实建议。</li>
</ul>
</div>
<div class="sec">
<h2>📎 来源汇总</h2>
<div class="foot-note">本期共采集 N 条信息,采用 M 条。数据来源:newsnow 热点聚合、bocha_search 联网检索、竞品档案库。所有外部信息仅供参考,重大决策请与内部经营数据交叉验证。</div>
</div>
</div>
</body>
</html>
<!-- REPORT_HTML_END -->
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 177 lines · 52 tokens per session scan A b43cc0ce746b
market-daily-brief is a skill published in the GitHub repository zj-unicom-ai/UniEmployee (362 stars, last pushed 8d ago), licensed MIT. It adds 52 tokens to every session and 2,974 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-13.
Other skills, from other repositories
five-ring-ontology-engineering-lifecycle
A five-stage method for planning an enterprise ontology, a structured model of the concepts and relationships used across a business, from source material through ongoing use.
ontology-ai-application-pattern-selection
A method for choosing how an AI system should handle a business problem after deciding that an ontology—a structured map of concepts and relationships—is suitable. It compares patterns such as workflow automation, decision support, knowledge answering, collaboration, and continuous planning.
ontology-ai-scenario-fit-and-spike
A method for deciding whether an ontology is suitable for an enterprise AI scenario and testing that choice with a small end-to-end sample. An ontology is a structured model of concepts, rules, and relationships.
ontology-constraint-and-knowledge-injection
A method for deciding how an ontology’s knowledge should reach an AI model: directly in its instructions, through RAG, or through fine-tuning. RAG retrieves relevant information at answer time; fine-tuning changes the model using training examples.
ontology-golden-case-testing
A testing method for an ontology, which is a structured model of business concepts and rules. It starts with real business questions and expected answers, then adds boundary, missing-data, conflict, permission, and regression tests.
ontology-model-multilayer-quality-gate
A quality-review process for an ontology—a structured map of concepts and relationships—created by an AI model. It combines independent review, software checks, and business-expert checks before production use.