How this skill is triggered — by the user, by Claude, or both
Slash command
/plugin-ops:diagnoseThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
---
Investigate and report plugin defects. Frame findings as plugin defects, not agent failure — if a skill wasn't triggered, that's a description quality issue; if a hook failed, that's a configuration defect.
Parse $ARGUMENTS:
--skill SKILL-NAME: Focus on a specific skill.If no plugin found at path, ask the user which plugin to diagnose.
Cache guard: ~/.claude/plugins/cache/ is READ-ONLY. Resolve to the real git repo via atlas: atlas_search_projects(query="plugin-name").
.claude-plugin/plugin.json exists, valid JSON, has name/version/descriptionskills/, hooks/, agents/, knowledge/) existFor each skill (or --skill target):
Structure: Valid YAML frontmatter with name + description, valid markdown body.
Description quality (the #1 reason skills don't trigger):
Content: No bare $VARIABLE outside code blocks, no broken refs to missing knowledge/files.
hooks/hooks.json valid JSON with correct event namesgh issue list -R {repo}claude plugin listPlugin Diagnostics: {plugin-name} v{version}
{DEFECT-1}: {category} — {short title}
Location: {file-path}:{line-number}
Severity: critical | major | minor
Details: {what's wrong and why}
Fix: {concrete change}
Summary: {N} defects, root cause, fix priority
If skill description quality is the root cause, recommend running /skill-creator (from the plugin-dev plugin) to iterate with test prompts and description optimization.
npx claudepluginhub ivintik/private-claude-marketplace --plugin plugin-opsValidates Claude Code plugins for compliance with official guidelines including plugin.json manifest, directory structure, and formats for agents/skills/commands to prevent installation failures.
Validates Claude Code plugin directories against Anthropic spec via structural checks (plugin.json, files, permissions) and content grading (SKILL.md, commands). Use for building, auditing, or marketplace prep.
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.