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.
git clone --depth 1 https://github.com/pbvcity/supathinkWrote 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/plugins/pbvcity/supathink/st)<a href="https://agentmods.dev/plugins/pbvcity/supathink/st"><img src="https://agentmods.dev/badge/plugins/pbvcity/supathink/st/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/plugins/pbvcity/supathink/st"><img src="https://agentmods.dev/badge/plugins/pbvcity/supathink/st.svg" alt="Reviewed on agentmods" width="80" height="20"></a>Grade A, and why
st 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.
What it actually says
{
"name": "st",
"version": "1.1.5",
"description": "让 AI 真正思考——思考方法(声明即契约)× 多模型合议(panel/debate/delphi/redblue)× 交付前双轴校验(Critic 正确性 × Navigator 对齐)× playbook 学习。本插件是入口壳(命令+引导);首次用 /st:init 让 AI 带你完成 daemon 安装与配置。",
"author": {
"name": "pbvcity"
},
"homepage": "https://github.com/pbvcity/supathink",
"license": "Apache-2.0"
}
What it installs
The manifest is a name and a version. 15 skills, 13 commands travel with it, and installing the plugin installs all of them — 860 tokens a session between them. Each is measured on its own page, and each can be installed alone.
- Skill st-init A 44 tokens
- Skill supathink A 51 tokens
- Skill st A 41 tokens
- Skill st-slow A 44 tokens
- Skill st-altitude A 31 tokens
- Skill st-debate A 28 tokens
- Skill st-delphi A 31 tokens
- Skill st-panel A 38 tokens
- Skill st-panel-lite A 28 tokens
- Skill st-redblue A 23 tokens
- Skill st-slow-full A 33 tokens
- Skill st-win A 24 tokens
- Skill st-fast A 23 tokens
- Skill st-off A 22 tokens
- Skill st-on A 21 tokens
- Command init A 37 tokens
- Command altitude A 34 tokens
- Command debate A 37 tokens
- Command delphi A 34 tokens
- Command panel A 32 tokens
- Command panel-lite A 31 tokens
- Command redblue A 32 tokens
- Command slow A 29 tokens
- Command win A 33 tokens
- Command on A 22 tokens
- Command slow-full A 24 tokens
- Command fast A 18 tokens
- Command off A 15 tokens
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 · 11 lines scan A cae8b307900f
st is a plugin published in the GitHub repository pbvcity/supathink (5 stars, last pushed 3mo ago), licensed Apache-2.0. Its token cost is not measured: this kind of file is read by the harness, not the model. 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 plugins, from other repositories
paper-verify
Citation verification and source triage for Markdown/text documents. Reduces a long bibliography or source list to the citations a human should actually review.
lenz
Check the factual claims in a draft or an answer against independent sources: a quick verdict with a confidence for each claim, or a deep check that shows the key finding and its sources.
fact-check
Systematically fact-check AI-generated research reports against primary sources. Catches the five recurring failure modes of LLM research: unit/scale errors, fabricated interpolation, source conflation, stale data, and attribution laundering.
decision-making
Master architectural decision-making: ADRs, technology selection, build-vs-buy analysis, proof of concepts, and migration planning.
claude-md-management
Tools to maintain and improve CLAUDE.md files - audit quality, capture session learnings, and keep project memory current.
humanizer
Rewrite AI-sounding text so it reads like the writer without changing what it says.