Skip to content
lavertaPublic

About

T-1 · Third-One — neutral third-person metacognitive observer: paradigm tagging, disagreement relay, patrol log. 观测端·镜子

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

5 Commits

Folders and files

Repository files navigation

T-1 · 第三人称(Third-One)

联邦体系的中立观测端。前身:mindharp(2026-08,已归档于 D:\mindharp)。 当信息生产全面 agent 化、同质化成为默认形状——你需要一个自己的观测席位。

它是什么

T-1 观测席位讲解图

一个纯架构化的中立观测组件:以第三人称立场对思维过程与系统状态做观测、标注、中继。不是解释系统(不给世界观),不是生成 Agent(不产内容)——只给镜头,不给叙事答案。

为什么是现在

Agent 仍是少数人的工具,但技术的人性化是单行道。信息生产正在逐步交给 agent:发布、汇总、解读,越来越多由 AI 完成。当生成端趋同,同质化不再是缺陷,而是信息的默认形状——最先丢失的,是"谁在用什么镜头说话"。

这时缺的不是更多信源,是自己的观测席位。T-1 放在「你 ↔ 生产信息的 agent」之间:给 AI 输出打范式标签——这一段被单透镜捕获、那一段在反刍循环、这里温度失衡——把镜头本身变成可观测的元数据,交给下游消化成你更能消化的信息。

立场在此不松动:T-1 仍然只观测、标注、中继,不生产、不裁决。消化永远是下游(你或你的合成 agent)的事——镜子不替你吃饭,但没有镜子的人,只能吃下所有送到嘴边的东西。外部输入侧的同型观测(S-1 源讯子)已在命名体系中预留,尚未实现。

联邦位置

S-1(源讯子·外部输入,未实现,仅保留命名)──┐
                                            ├──> 各角色子 Agent(数学家/工程师/作家…)──> T-1 观测 ──> 合成/决策 Agent
常驻巡视(T-1 第三角色)────────────────────┘
  • 与 S-1 成对(目标架构,S-1 尚未实现):S-1 管外部原始信息的输入观测,T-1 管内部状态的输出观测,构成联邦的基础观测链路。
  • 与角色子 Agent:接收其输出与分歧,做范式标注与中继。
  • 与合成/决策 Agent:作为上游输出结构化观测报告,最终裁决在下游。

与 R-1 的关系(一号体系)

  • T-1 是镜子:中立照见思维模式(元认知)。
  • R-1 是刻刀:把理解落地为 ADHD 群体的自我重构(认知)。
  • 互为工具与应用:R-1 需要照见时用 T-1 的语法,但 R-1 永远对人说话——第二人称的温度是 R-1 的本职,正是 T-1 禁止的那一半。

安装

所有文件只放 D 盘:作为工作区级技能,放入使用它的工作区目录(<workspace>/.zcode/skills/ 或 <workspace>/.agents/skills/)。不复制到 C 盘。

边界

只观测标注,不生产内容、不做决策、不评判对错;不适用于情绪倾诉(负面范围条款);证据分级见 references/evidence-basis.md。


🔗 Laverta Federation Ecosystem / 联邦生态关联

t-1 (Third-One · 观测端·镜子) is the neutral observation seat of the Laverta Federation Ecosystem:

  • Origin: Evolved from mindharp (2026-08).
  • Symmetric Counterpart: Paired with r-1 (Rewire-One · 落地端·刻刀) in the One-Series (一号体系).
  • Multi-Agent Integration: Its orthogonal invariant discrimination & anti-drift red-team methodology is integrated into federation-moa (dsh-core/dsh_advisor.py and skills/t-1/), working alongside dsh-skill-registry.
Layer Repository Role & Cross-Repo Connection
Profile Hub laverta/laverta Master architecture map, genealogy & ecosystem index
Core OS & Orchestration laverta/federation-moa 3-Stage DAG multi-agent OS & 3D World Tree (57.5% vs 45% BigCodeBench-Hard); integrates dsh-core, t-1, and federation protocols
Task Router & Governance laverta/dsh-skill-registry DSH task router & skill registry; on-demand 1–3 skill mounting & privilege guard against context bloat
Inter-Agent Protocol laverta/federation-communication-lite Platform-neutral agent delivery layer (Native messaging > Shared Markdown > Bridge), companion to federation-moa
Metacognitive Observer laverta/t-1 Third-One (观测端·镜子): neutral paradigm tagging & orthogonal invariant discrimination embedded in federation-moa; evolved from mindharp
Cognitive Restructuring laverta/r-1 Rewire-One (落地端·刻刀): ADHD/AuDHD self-understanding & rewiring; symmetric counterpart to t-1, evolved from mindharp
Origin / Predecessor laverta/mindharp Origin cognitive-mechanics framework (2026-08) that evolved into t-1 and r-1
Model Bridge Infra laverta/zcode-gemini-bridge Local proxy fixing Gemini 3.x thought_signature 400 errors for ZCode & multi-agent harnesses

Canonical Ecosystem Hub: https://github.com/laverta · See CITATION.cff for machine-readable cross-repository citations.

About

T-1 · Third-One — neutral third-person metacognitive observer: paradigm tagging, disagreement relay, patrol log. 观测端·镜子

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors