cognitive-input-classifier

cognitive-input-classifier is a skill for Cursor from TashanGKD/cognitive-os. It costs 113 tokens per session (1,878 once invoked), scanned A, a copy of cognitive-input-classifier, MIT.

An input-routing classifier decides whether a message is a knowledge note or a task to carry out, using words found in that message.

In plain words
What is it for?
Use it to sort requests before work begins, such as separating an insight to record from a request to build, fix, analyse, or deploy. It does not start the next skill automatically.
Why use it?
It removes guesswork when a request could be interpreted in different ways. It gives a clear route and reason, or asks one question when the type cannot be determined.

Skill for Cursor

Written for Cursor: installed under .cursor/.

Good fit Use it to sort requests before work begins, such as separating an…

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/tashangkd/cognitive-os/cognitive-input-classifier
Install

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.

Any agent
npx skills add TashanGKD/cognitive-os --skill cognitive-input-classifier
Clone the repo
git clone --depth 1 https://github.com/TashanGKD/cognitive-os

Made for: Cursor.

Wrote 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.

agentmods badge for cognitive-input-classifier

README.md
[![agentmods](https://agentmods.dev/badge/skills/tashangkd/cognitive-os/cognitive-input-classifier.svg)](https://agentmods.dev/skills/tashangkd/cognitive-os/cognitive-input-classifier)
Your own site
<a href="https://agentmods.dev/skills/tashangkd/cognitive-os/cognitive-input-classifier"><img src="https://agentmods.dev/badge/skills/tashangkd/cognitive-os/cognitive-input-classifier.svg" alt="Measured on agentmods" height="20"></a>
Per session 113 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,878 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin 100% copy Near-identical to another mod in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00113 $0.01878
Opus 5 $0.00056 $0.00939
Sonnet 5 $0.00023 $0.00376
Haiku 4.5 $0.00011 $0.00188

Measured 6d ago against content hash 13643608f62b, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

cognitive-input-classifier 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 6d 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.

Origin

This is a copy

100% identical to cognitive-input-classifier — 0 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.

.cursor/skills/cognitive-input-classifier/SKILL.md · 136 lines

How it starts

The opening of the file, as written. The whole thing — 136 lines — stays where its author put it; the contents beside it link to each section on GitHub.

认知输入分类器(cognitive-input-classifier)

强绑定 Rule:R2 NO_FABRICATION(判断理由必须基于输入文本中实际存在的信号词) R13 RULE_BEATS_STYLE(准确判断优先于表述流畅)


核心原则

判断理由必须基于文本证据:从用户输入中找到具体信号词,引用它来支撑结论。不允许「感觉上像认知更新」这类无证据判断。

混合时强制拆分:同时发现两类信号词时,必须走「AB混合」分支,不允许取占比大的一类压过另一类。

无法判断时只有两个出口:追问一次,或默认路径B并告知可纠正。不允许输出「这是一个既有认知又有任务性质的输入」类模糊结论。


激活后立即执行

Step 1  分析输入的意图信号
        从用户输入中(不是从上下文,而是从当前这段文字本身)找信号词:
        
        认知信号词(指向路径A):
          「我发现」「我觉得」「我想到」「记录一下」「碎片」「洞见」
          「理解X是什么」「X的规律是」「这让我想到」「我注意到」
          「想法」「感悟」「我有个看法」
          「先记录」「先按规范记录」「记录一下这个洞见/方向/原则/发现」
        
        ⚠️ CO-BUILD-LOG 消歧:「先按规范记录」+「洞见/方向/规律/原则/新发现」
           → 路径A(cognitive-capture-fragment,写 L2 碎片),不是 CO-BUILD-LOG
           CO-BUILD-LOG 仅用于「记录当前正在执行的建设任务的过程决策」,
           而非「记录从过程中产生的知识性洞见结果」。
        
        任务信号词(指向路径B):
          「帮我做」「写X」「执行」「实现」「修复」「部署」「设计」
          「新建」「创建」「分析[X并输出]」「整合[X到某处]」
          祈使句动词开头且带具体可交付产物的描述
        
        ⚠️ 优先基于当前输入文字中出现的信号词做判断。
           若信号词指向结论,不用上下文语义来覆盖文字证据。

Step 2  判断并输出路由结论(格式固定)
        
        情形1:只发现认知信号词,无任务信号词
          → 输出:
          「📗 路径A:认知更新。
            建议触发:cognitive-capture-fragment
            判断依据:输入中的「[引用实际词句]」表明这是一个认知洞见/思考记录。」
        
        情形2:只发现任务信号词,无认知信号词
          → 输出:
          「📋 路径B:任务执行。
            建议触发:[基于输入的动词和领域推断对应Skill;不确定时写「[请指定Skill]」]
            判断依据:输入中的「[引用实际词句]」表明这是一个需要执行的任务。」
        
        情形3:同时发现两类信号词(AB混合)
          → 输出:
          「🔀 路径AB混合。
            认知部分:「[引用内容]」→ 路径A → cognitive-capture-fragment
            任务部分:「[引用内容]」→ 路径B → [对应Skill]
            请问先处理哪个?」
        
        情形4:无法找到明确信号词
          → 追问(最多1次):
          「❓ 无法从文字中找到明确分类信号。请问这个输入的核心目的是:
            (A) 理解/记录某件事(认知更新)
            (B) 做成某件事(任务执行)?」
          → 若用户回答 → 按回答输出对应情形1或2的结论
          → 若用户仍不确定 → 默认路径B:「默认路径B(任务执行),如有误请直接纠正」

Step 3  输出结论后强制停止,等待用户指令
        
        ⚠️ 关键约束:输出路由结论后,本 Skill 执行完成。
           不继续执行任何其他步骤,不触发任何下游 Skill。
           即使上下文中有明显的「下一步」,也不主动执行。
        
        等待用户的明确指令:「确认」「走A」「走B」「用路径X触发」等。
        
        若用户不认可结论:
        → 无条件接受纠正,输出:「收到,更正为路径[X],你可以直接触发 [对应Skill]」
        → 不辩解,不再次分析

Read the full file on GitHub · 136 lines

Changes

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.

  1. 6d ago First seen · 136 lines · 113 tokens per session scan A 13643608f62b

Subscribe to this mod's changes

cognitive-input-classifier is a skill published in the GitHub repository TashanGKD/cognitive-os (8 stars, last pushed 5mo ago), licensed MIT. It adds 113 tokens to every session and 1,878 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. It is 100% identical to cognitive-input-classifier, differing in 0 lines, and is treated as a copy.