From ynh
Guide graduation from a personal harness to a team setup with delegation. Creates a team harness that delegates to personal harnesses.
How this skill is triggered — by the user, by Claude, or both
Slash command
/ynh:ynh-team-setupThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are guiding a user through creating a team harness that delegates to personal harnesses. This is the "graduation" from individual use to team-wide adoption.
You are guiding a user through creating a team harness that delegates to personal harnesses. This is the "graduation" from individual use to team-wide adoption.
Read these references to understand delegation and Git URL formats:
references/delegation.md for delegates_to syntax, Git URL formats, auth, and vendor supporttestdata/team-harness/.harness.json for the team harness structureAsk the user:
Explain how ynh delegation works (see references/delegation.md):
Delegation means a team harness knows about personal harnesses. When someone runs the team harness and asks for something that a personal harness handles, the AI vendor can delegate to it as a subagent.
Two modes of use:
team-dev "deploy checklist" and it delegates to the right specialistdavid directly when you need your full personal contextThe team harness bundles shared standards (rules, skills) while individual harnesses carry personal preferences.
Ask where to create the team harness. Options:
Suggest a name like team-dev or <company>-dev.
Ask what shared artifacts the team needs:
They might also want to pull from external repos via includes.
Create the team harness directory.
.harness.json:
{
"$schema": "https://eyelock.github.io/ynh/schema/harness.schema.json",
"name": "<team-name>",
"version": "0.1.0",
"description": "<team description>",
"default_vendor": "<vendor>",
"delegates_to": [
{"git": "<personal-persona-git-url>"}
]
}
Git URL format for delegates_to - see references/delegation.md for the three formats:
github.com/user/harness (expands to SSH)[email protected]:user/harness.githttps://github.com/user/harness.gitIf the personal harness isn't in Git yet, explain they'll need to push it first for delegation to work. Show them how:
cd <harness-dir>
git init && git add . && git commit -m "Initial harness"
# Push to their Git hosting
Generate any shared artifacts they requested (rules, skills, etc.) following standard formats (skills need SKILL.md with frontmatter, agents need markdown with frontmatter, rules and commands are plain markdown).
Show the team installation flow:
# Team member installs the team harness
ynh install <team-harness-git-url>
team-dev # interactive session with team config
# They can also install their own personal harness
ynh install <personal-persona-git-url>
david # personal session
Explain the vendor standardization: setting default_vendor in .harness.json ensures everyone uses the same AI vendor, but individuals can override with -v.
If any repos are private, explain the auth model (see references/delegation.md). Key points:
[email protected]:...) recommended for private reposgit binary - if git clone works, ynh worksAfter the team harness is working:
ref: v1.0.0 in includes for stable referencespath field in delegates_to (see references/delegation.md).npx claudepluginhub eyelock/ynhProvides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.