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 agentmods add agents/supsub-ai/supsub-cli/phrasinggit clone --depth 1 https://github.com/SupSub-AI/supsub-cliWhat 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 | $0.00000 | $0.01034 |
| Opus 5 | $0.00000 | $0.00517 |
| Sonnet 5 | $0.00000 | $0.00207 |
| Haiku 4.5 | $0.00000 | $0.00103 |
Grade A, and why
phrasing 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 2d 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.
What it actually says
怎么说更准
大部分时候随便说就行。但有几种说法天生有两种理解，助手可能会猜错。这里把它们列出来，顺便给个更清楚的说法。
「搜一下 X」还是「X 最近怎么样」
这是最容易理解偏的一组。
| 你想要的 | 这么说 |
|---|---|
| 在全站找跟这个话题有关的文章 | 「搜一下 AI 安全」「帮我找找讲 AI 安全的」 |
| 看你已经订阅的某个号最近发了什么 | 「看看『阮一峰』最近发了什么」 |
差别在动词。「搜」是去全站翻找，「看看最近」是打开你自己订阅的那个号。
如果你说「看看 Anthropic 的内容」而你正好订阅了 Anthropic 的博客，助手应该先给你看你订阅的那个，而不是去搜一堆别人写的评论文章。要是它搞错了，补一句「我是说我订阅的那个」就行。
「搜 X」还是「找 X 这个号」
| 你想要的 | 这么说 |
|---|---|
| 找文章 | 「搜一下大模型」「找找讲 RAG 的内容」 |
| 找一个号，准备订阅 | 「找**『机器之心』这个公众号**」「我想订阅少数派」 |
想找号的时候，把「公众号」「这个号」「想订阅」这类词说出来，助手就不会去翻文章了。
::: tip 话题词不等于号名 「大模型」「RAG」「AI agent」这种词，助手默认当成文章话题去搜，这通常也是你要的。 :::
「订阅」「关注」「分组」别混着说
这三个词在 SupSub 里指三样不同的东西，说错了助手会去查错的地方：
| 说法 | 指的是 |
|---|---|
| 我订阅了哪些 | 你追的那些号 |
| 我关注了哪些 / 我的关注点 | 你设的主题方向 |
| 我的分组 / 归到…里 | 你建的文件夹 |
举个会出错的例子：你想看自己追了哪些号，但说成「我关注了哪些」，助手会去给你列主题方向。改说「我订阅了哪些号」就准了。
「最近」不等于「没看的」
| 你想要的 | 这么说 |
|---|---|
| 不管看没看过，按时间看最近的 | 「这个号最近发了什么」 |
| 只看还没读的 | 「这个号还有什么没看的」 |
这个区别在你已经把某个号读完的时候特别明显——问「还有什么没看的」会得到「没有了」，但问「最近发了什么」照样能看到内容。
说「X」的时候当心
字母 X 现在既是推特的名字，又常被拿来当占位符。
- 想订阅推特账号：说「订阅推特上的 xxx」或「X 平台上的 xxx」，把平台说清楚
- 只是随口举例：助手不该把它当成推特
一般不会出问题，但如果你发现助手莫名其妙去找推特账号了，把话说全就行。
让助手标记已读时
标成读过之后没法反悔，所以说清楚范围很重要：
| 你想要的 | 这么说 |
|---|---|
| 某个号全部标成读过 | 「『量子位』这个号我读完了」 |
| 某个主题方向全部标成读过 | 「『AI 动态』这个关注点我都看过了」 |
助手动手前会跟你确认一次。不确定的时候可以先说「先列出来我看看」，确认没问题再让它清。
::: warning 只有整个号、整个关注点两种 没法只标某一篇，也没法整个分组一起标。想清一个分组，助手会把里面的号列出来，一个一个跟你确认。 :::
它理解偏了怎么办
直接说就行，不用重开一句：
不对，我是说我订阅的那个
我要的是文章，不是公众号
只要没看过的
先别标，让我看看列表
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.
- 2d ago First seen · 87 lines · 0 tokens per session scan A e1f04677a99b
phrasing is an agent published in the GitHub repository SupSub-AI/supsub-cli (5 stars, last pushed 1mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,034 tokens. 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 2026-08-31.
Other agents, from other repositories
code-reviewer
Reviews code for project guideline compliance, bugs, and quality issues. Use after writing code, before commits, or before PRs. Specify files to review or defaults to unstaged git changes. High-confidence issues only (80+) to minimize noise.
docs-impact
Reviews documentation affected by code changes. Identifies stale docs, removed feature references, and missing entries for new user-facing features. Reports findings with specific fixes. Advisory only - does not modify files.
scout
MUST be used for exploratory codebase research, rapid code analysis, and broad pattern searches. Fast read-only scout returning compressed context for handoff.
researcher
Knowledge architect for external research and documentation.
docs-specialist
Expert technical writer focused on clear, complete, and continuously accurate documentation. Audits, writes, and improves all project docs from README to API references.
triage-scan
You are a triage analyst. ./input.json names one repo and the exact source tree to read it against.