Plugin Ecosystem

abstract

Meta-skills infrastructure — skill authoring, hook development, modular design patterns, and evaluation frameworks.

New

Meta-skills infrastructure for the Claude Code plugin ecosystem — skill authoring, hook development, modular design patterns, and evaluation frameworks.

Source: claude-night-market marketplace Version: 1.4.3 Author: Alex Thola

Key Commands

CommandDescriptionWhen to Use
/create-skillAuthor a new Claude Code skillWhen building a new skill from scratch
/create-commandCreate a new slash commandWhen adding a command to a plugin
/create-hookDevelop a new hookWhen adding lifecycle hooks
/test-skillTest a skill in isolationAfter creating or modifying a skill
/bulletproof-skillHarden a skill against edge casesWhen preparing a skill for production
/evaluate-skillEvaluate skill qualityWhen assessing a single skill
/skills-evalEvaluate all skills in a pluginWhen auditing a plugin's skill quality
/hooks-evalEvaluate all hooks in a pluginWhen auditing hook quality
/analyze-skillDeep analysis of a skillWhen investigating skill behavior
/analyze-hookDeep analysis of a hookWhen investigating hook behavior
/validate-pluginValidate plugin structureWhen checking plugin.json and structure
/validate-hookValidate hook configurationWhen checking hook config
/plugin-reviewFull plugin quality reviewWhen reviewing an entire plugin
/improve-skillsSuggest skill improvementsWhen looking for skill optimization opportunities
/make-dogfoodGenerate dogfooding scenariosWhen creating test scenarios for skills
/aggregate-logsAggregate skill execution logsWhen analyzing skill usage patterns
/context-reportReport on context usageWhen analyzing context window usage
/estimate-tokensEstimate token consumptionWhen planning token budgets

Skills

SkillWhen It Activates
escalation-governanceWhen escalation decisions arise that need governance
hook-authoringWhen creating or modifying hooks
hooks-evalDuring hook evaluation and quality assessment
makefile-dogfooderWhen dogfooding Makefile-based workflows
methodology-curatorWhen curating and organizing methodologies
modular-skillsWhen designing skills with modular architecture
performance-optimizationDuring performance optimization of skills
shared-patternsWhen applying cross-plugin shared patterns
skill-authoringWhen creating or modifying skills
skills-evalDuring skill evaluation and quality assessment
subagent-testingWhen testing subagent behavior

Agents

AgentPurpose
meta-architectDesign plugin architecture and skill organization
plugin-validatorValidate plugin quality against best practices
skill-auditorAudit skill implementations for issues
skill-evaluatorEvaluate skill effectiveness with metrics
skill-improverSuggest and apply skill improvements

Common Workflows

Build a New Skill

The standard flow for creating a production-quality skill:

  1. Create — Run /create-skill and describe what the skill should do. Abstract scaffolds the skill file with the correct structure and metadata.
  2. Test — Run /test-skill to test the skill in isolation, verifying it activates correctly and produces expected results.
  3. Harden — Run /bulletproof-skill to test edge cases, add defensive guards, and ensure the skill handles unexpected inputs gracefully.
  4. Evaluate — Run /evaluate-skill to get a quality score and improvement suggestions.

Plugin Quality Check

When auditing an entire plugin for quality:

  1. Review — Run /plugin-review for a comprehensive quality assessment covering structure, skills, commands, hooks, and agents.
  2. Evaluate skills — Run /skills-eval to evaluate every skill in the plugin, producing a report with scores and improvement priorities.
  3. Improve — Run /improve-skills to get specific, actionable improvement suggestions for each skill that scored below threshold.

Tips

  • Abstract is a meta-plugin — it is used to build and improve other plugins, not for day-to-day development work.
  • The skill-authoring skill activates automatically when you are editing SKILL.md files, even without running a command.
  • /estimate-tokens is useful for keeping skills within token budgets. Large skills consume context window space.
  • The modular-skills pattern encourages breaking large skills into composable pieces.

See Also