From ynh
Guided walkthrough for using ynd inspect to bootstrap a project's AI skills and agents from its codebase signals.
How this skill is triggered — by the user, by Claude, or both
Slash command
/ynh:ynd-inspectThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are guiding a user through using `ynd inspect` to analyze their codebase and generate tailored skills and agents.
You are guiding a user through using ynd inspect to analyze their codebase and generate tailored skills and agents.
Read these references to understand what inspect does and the artifact formats it generates:
references/inspect-workflow.md - How inspect works, output directories, vendor-specific pathsConfirm the user has:
ynd built and on PATH (or they know the path to the binary)claude, codex, or cursor)If they're missing the LLM CLI, explain that inspect needs one to analyze the codebase. Point them to installation docs.
Ask which project they want to inspect. They should cd into the project root. Explain that inspect looks for signal files (build configs, test configs, CI files, etc.) to understand the project.
Suggest starting with interactive mode to review the analysis:
ynd inspect
Walk them through each step:
Help them evaluate each proposal:
Encourage them to skip proposals that feel generic. The best skills automate workflows they already do manually.
Explain the output options:
.{vendor}/ (e.g., .claude/skills/, .claude/agents/). This is where the vendor CLI looks for them.-o .: write to project root (skills/, agents/). Use this if the project itself is a harness/plugin.-o /path: write to any custom directory.Once they're happy with the proposals, generate:
# Interactive — review each one
ynd inspect
# Or auto-generate all at once
ynd inspect -y
After generation, validate and lint the generated artifacts:
ynd lint .claude/skills/ .claude/agents/
Check that frontmatter is correct and the content is specific to their project.
Explain that inspect is meant to be run periodically as the project evolves. New dependencies, new test frameworks, or new CI pipelines will produce different suggestions. They can re-run and update existing artifacts.
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.