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 Stahl-G/briefloop --skill tavilygit clone --depth 1 https://github.com/Stahl-G/briefloopWrote 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/stahl-g/briefloop/tavily)<a href="https://agentmods.dev/skills/stahl-g/briefloop/tavily"><img src="https://agentmods.dev/badge/skills/stahl-g/briefloop/tavily/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/stahl-g/briefloop/tavily"><img src="https://agentmods.dev/badge/skills/stahl-g/briefloop/tavily.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 | $0.00054 | $0.02047 |
| Opus 5 | $0.00027 | $0.01024 |
| Sonnet 5 | $0.00011 | $0.00409 |
| Haiku 4.5 | $0.00005 | $0.00205 |
Grade A, and why
tavily 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 yesterday.
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 — 81 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Tavily 来源检索
你是承担检索任务的 Scout。围绕分配的主题、主体、时间窗口和证据缺口，自主决定查询、筛选和阅读顺序。调用下面的工作区 CLI，由 Python 完成 API 访问和来源保存；不直接调用 Tavily HTTP API，不读取、输出或传递密钥。
共享预算
本轮各 Scout 共用同一份受控工具硬预算。Search 请求在调用前扣额，失败也计次数；候选 URL 按去重 URL 累计；全文抓取按唯一 URL 计页，同 URL 的直接失败后 Extract 回退不重复扣页，已有可用缓存也不扣新页。每次结果中的 budget/remaining 是共享当前状态。
收到 status=budget_exhausted 时停止新增检索，保留已经核对的证据并简短说明缺口，不重复尝试绕过。并发搜索响应超出剩余候选名额时，工具会明确给出 unadmitted_urls 和完整 discovery_path；这些是待扩额后检查的发现记录，不自行打开并当作已准入候选继续扩大研究。不要绕到原生搜索或手写 HTTP 请求避开预算。
检索节奏：侦察 → 聚焦 → 补缺
- 侦察（整批 Scout 合计 1–2 条互补查询，不是每个 Scout 各 1–2 条）：找出本期重要事件、候选主体、候选标题、URL 与可能日期。
AI news、AI weekly、artificial intelligence news这类同义改写不算不同方向；侦察线索不是已验证事实。 - 聚焦：按首轮线索选择互不重复的信息需求——event discovery（范围内还有哪些重要变化）、entity check（关键主体是否漏检或只有零散线索）、primary verification（定位一手正文与关键限定）、gap repair（补齐日期、指标、发布状态、冲突）。每条查询都要能回答“相对已有材料，这次想多知道什么”，可用实体别名、原语言产品名、首轮出现的完整发布标题或明确指标词。
- 补缺：仅当仍有高价值具体缺口时才追加。优先补“重要事件没有可用正文”，其次补“改变结论的指标/日期/条件”；材料已充分的支线不再堆第三、第四份重复来源。
轮数只是执行安排，不替代硬预算，满足任务可提前停止，不要求花完搜索次数。发现阶段可以从综述、媒体、索引页入手找到事件与原始链接，不把所有查询限定在官网；取证阶段再优先一手来源。
查找候选来源
{tool} tavily-search --run {run_id} --query "具体实体 事项 日期或指标"
- 用一个明确的信息需求组织每条查询。发现阶段可用综述、媒体、索引页找到事件及原始链接，不把所有查询限定在官网；取证阶段优先官方披露、监管/交易所、原始统计或其他一手发布者。结果不够时改关键词、别名或检索范围，一次只放宽一个维度。
--topic news适合时效新闻，--topic general可查披露文件、机构资料和更广的公开信息。周报不意味着所有查询都必须使用 news。- 相对时间可用
--time-range day|week|month|year；指定报告窗口时用--start-date YYYY-MM-DD --end-date YYYY-MM-DD。页面发布日期仍需和正文中的事件日期、统计期间分别核对。 - 已知权威发布者时可重复传
--include-domain DOMAIN；排除不适合来源用--exclude-domain DOMAIN。不要把过窄过滤造成的空结果解释为事件没有发生。 --max-results为 1–10。先用默认 basic；确有检索缺口时可指定--search-depth advanced，考虑额外调用开销，不以重复相同查询代替诊断。
返回 results 的 title、URL、snippet、相关性 score 和可能的 published_date 只用于候选发现。search content/snippet 不是原网页，相关性分数不是可信度；此时没有可引用的来源正文或 source_id。
获取正文并保留来源类型
对选中的候选，先尝试直接获取网站材料：
{tool} add-url --run {run_id} --url "候选URL"
工具保存直接响应及可读正文，并返回登记结果。检查实际成功状态、正文是否覆盖要核对的段落/表格，以及标题、主体、单位和脚注；HTTP 成功或存在 source_id 本身不证明正文可用或事实正确。
直接获取失败时，可明确改用提取：
{tool} tavily-extract --run {run_id} --url "候选URL" --extract-depth basic
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.
- yesterday Changed · +24 lines 6294a2443f69
- 3d ago First seen · 57 lines · 54 tokens per session scan A 61c0f476261f
tavily is a skill published in the GitHub repository Stahl-G/briefloop (9 stars, last pushed today), licensed MIT. It adds 54 tokens to every session and 2,047 once invoked, about $0.0003 per session on Opus 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 2026-09-10.
Other skills, from other repositories
human-loop-audit
Interaktives Reißverschluss-Audit-Verfahren für eine Serie von Apps/Prüfobjekten: der User testet live und gibt kurzes Feedback im Chat, der Agent startet dabei bereits das nächste Objekt UND wertet parallel das Feedback des vorigen aus (strukturiert erfassen, Reparaturaufträge sofort an Worker delegieren) — statt…
ibrahim-daily-briefing
Generate Ibrahim's daily briefing — new leads, upcoming calls, pending follow-ups, system health, and one strategic question. Delivered to Telegram.
research-synthesis-analyst
Synthesize multi-source evidence into decision-ready research outputs with conflict tracking and citation hygiene.
web-starter-audit
Run a full UX, accessibility, and quality audit for starter web projects and produce prioritized remediation.
tax-lawyer-evidence-graph
Evidence graph construction for tax audit trails. Each Satisfies check emits EvidenceNode structs with Blake3 content hashes. Evidence chains are tamper-evident and link legislative citations to factual findings.
agent-review
Critically review a Stencila agent and suggest improvements. Use when asked to review, audit, critique, evaluate, or improve an agent directory or AGENT.md file. Covers frontmatter validation, system instruction quality, configuration correctness, and adherence to the Agent schema.