Cotend init

Skill lhh1301506137/CoTend/skills/cotend-init

Use when the user wants one simple entry to initialize, update, repair, migrate, or resume the CoTend development framework for the current project, including migration from a legacy dual-ai project. Thin entry point that delegates to cotend-project-init Auto Mode and preserves pending-human-question boundaries before reporting whether it is safe to continue.From its SKILL.md

Install

npx -y skills add lhh1301506137/CoTend --skill cotend-init

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

2.0 KB, 386 tokens by cl100k_base, as published. Nobody here has run it

CoTend Init

This is the short visible entry point for the user's project workflow setup.

Use this when the user invokes /cotend-init, says "初始化开发框架", "更新开发框架", "修复开发框架", "接入 CoTend", "加载开发框架", asks to migrate a legacy dual-ai project, or asks whether the project can continue under the current framework.

Action

Load and follow cotend-project-init Auto Mode. That skill owns classification, Trellis active/dormant decisions, Plan Tree and understanding setup, decision/knowledge logs, risk and review boundaries, Code Context Harness, Acceptance/Test Harness, output shape, and framework-change evaluation routing.

If cotend-project-init is unavailable in the visible skill list, read the local Codex skill file from one of these fallback paths:

%USERPROFILE%\.codex\skills\cotend-project-init\SKILL.md
~/.codex/skills/cotend-project-init/SKILL.md

Then execute its Auto Mode exactly. Do not duplicate or reinterpret its detailed rules here.

Output Boundary

Report in the recorded project language, or English when none is recorded, using the cotend-project-init output contract:

Initialization Decision Summary
Your Confirmation Is Needed only when a human-only decision is pending
Detailed Initialization Audit only when the project-init report mode calls for it

If no human-only decision is needed, complete safe setup/update/repair and leave the project ready for the user to say continue. If a human-only decision is pending, make that explicit: a bare continue token, including 继续, does not answer the question unless the user provides a specific preauthorization.

What ships with it: 1 file

516 B alongside SKILL.md

agents/

openai.yaml516 B