YIKUAIBANZI

6 mods across 1 repository, 114 stars between them.

eval-consistency

01

YIKUAIBANZI/forge-skill

Skill Claude CodeCodex

A test for checking whether an AI consistently acts like a given persona across 10 conversation scenarios. It generates replies and scores them on five aspects, such as length, wording, interaction style, and rule-following.

114 4mo ago A 44 tokens original MIT

eval-debate

02

YIKUAIBANZI/forge-skill

Skill Claude CodeCodex

A Chinese-language test procedure for evaluating debates between simulated versions of the same person. It uses a persona and three decision scenarios, then produces a scored quality report.

114 4mo ago A 49 tokens original MIT

forge-persona

03

YIKUAIBANZI/forge-skill

Skill Claude CodeCodex

A persona-building workflow that studies someone’s messages, social posts, descriptions, and other material to create a profile of how they speak, interact, and make decisions.

114 4mo ago A 41 tokens original MIT

forge-self

04

YIKUAIBANZI/forge-skill

Skill Claude CodeCodex

A guided process for creating a digital version of a person's decision style and personality from conversation and optional personal materials. It records traits and preferences without trying to make the person seem better than they are.

114 4mo ago A 36 tokens original MIT

use-persona

05

YIKUAIBANZI/forge-skill

Skill Claude CodeCodex

A feature for speaking with the manner and habits of a chosen real person, based on that person’s stored profile.

114 4mo ago A 28 tokens original MIT

use-self

06

YIKUAIBANZI/forge-skill

Skill Claude CodeCodex

A decision exercise in which several versions of the user examine the same choice from different perspectives.

114 4mo ago A 36 tokens original MIT