Master Claude Code Plugin Mechanics with This Reference Skill

Where this fits

You're mid-sprint on a Claude Code plugin that keeps failing to load a hook payload, or you're about to author a new skill and need to know the exact manifest format. Instead of digging through source code or guessing at conventions, you need a single, authoritative reference that explains how plugins, skills, commands, agents, and hooks actually work in this repo. That's exactly the gap this skill fills—it gives you a domain pack that has been checked against shipped plugins, so you can move from confusion to working code faster.

Why agents benefit

• Self-serve definitions: The agent can pull up the exact meaning of "plugin," "skill," "command," and "hook" without leaving the conversation, reducing context-switching.
• Avoids ops confusion: The skill explicitly says not to use it for operations (that's night-market-operations), so the agent won't mix concerns and will route to the right tool.
• Debugging accuracy: It includes concrete examples from real plugins (imbue, abstract, herald), so the agent can compare against known-good implementations.
• Convention-aware authoring: When writing a new skill or command, the agent can follow the local conventions layered on top of standard Claude Code mechanics, ensuring consistency.
• Reduces hallucination: Because the reference is version-pinned (repo v1.9.15), the agent can trust that the details match the current codebase.

Practical scenarios

Debugging a hook payload mismatch

Your plugin's PreToolUse hook isn't receiving the expected fields. The agent loads this skill, checks the hook payload contract described in the reference, and quickly identifies that you're using an outdated field name from an older plugin version.

Authoring a new skill from scratch

You need to create skills/foo/SKILL.md with the right frontmatter and directory structure. The agent consults the reference to confirm the required metadata fields and the exact location, then scaffolds the file correctly the first time.

Understanding the marketplace registry

You're adding a new plugin to the night-market registry and need to know how the marketplace manifest maps to the plugin directory. The skill explains the registry format, so the agent can validate your changes against the expected schema.

Add it to your agent workflow

Install the skill directly from the night-market repository:

npx skills add athola/claude-night-market --skill claude-code-plugin-reference

Once installed, you can invoke it in a Claude Code session with a prompt like:

Use the claude-code-plugin-reference skill to explain the difference between a command and an agent, and show me a sample manifest from the plugins/imbue directory.

The agent will load the skill and return a precise, context-aware answer.

Compare before adopting

Before making this a permanent part of your toolchain, compare it against alternatives. Look at the quality signals: the reference is checked against shipped plugins on a specific date, which is a strong accuracy indicator. Check maintenance freshness—the repo is actively versioned (v1.9.15), so it's likely to stay current. Also consider whether you need a broader plugin reference or a more ops-focused skill; if your work is purely operational, this skill explicitly points elsewhere. Finally, assess workflow fit: if you're already using the night-market repo, this skill integrates naturally.

Why it is worth tracking

This skill stands out because it's not just a list of features—it's a curated, version-pinned domain pack that reflects real plugin implementations. With 335 stars and an active author, it has community momentum and is likely to evolve alongside Claude Code's plugin system. Track it if you're building or maintaining Claude Code plugins, and evaluate it again when the plugin API changes or you need to onboard new team members to your plugin architecture. For a deeper look, see the skill page on OpenAgentSkill.