Skills Market

Definitive standards for writing Skills Market plugins that help build paid Lucid Agents.

Core goals

1. Build paid agents with x402 and ERC-8004 flows.
2. Keep instructions concise and high signal.
3. Avoid repeating obvious CLI or language basics.
4. Make skills composable across agent workflows.
5. Use real, fetchable data instead of hardcoded mocks.

Required package structure

plugins/<skill-name>/
├── .claude-plugin/plugin.json
└── skills/
    ├── SKILL.md
    ├── references/   (optional)
    ├── examples/     (optional)
    └── scripts/      (optional)

SKILL.md contract

Every skill must include YAML frontmatter followed by markdown instructions.

---
name: my-skill
description: "Describes behavior. Use when [trigger condition]."
---

Description rules

1. Write in third person, not imperative voice.
2. Include an explicit Use when ... trigger.
3. Include concrete terms users will naturally ask for.
4. Keep under 1024 characters.
5. Quote long descriptions or any text containing : to avoid YAML parse errors.

Content limits

• Keep SKILL.md under 500 lines.
• Put detailed guidance in skills/references/*.md.
• Keep references one level deep from SKILL.md (no nested reference chains).
• Any reference file over 100 lines must include ## Contents at the top.

.claude-plugin/plugin.json contract

Required fields:

Field	Type	Requirement
name	string	Must match directory name, lowercase with hyphens
description	string	One-line plugin purpose
version	string	Semver (for example 1.0.0)
author	object	Must be { "name": "...", "email": "..." }
license	string	SPDX-like identifier (for example MIT)
keywords	array	Must include lucid-agents
skills	string	Must be "./skills"

Example:

{
  "name": "my-skill",
  "description": "Generates payment schemas for x402 agents",
  "version": "1.0.0",
  "author": { "name": "Jane Doe", "email": "jane@example.com" },
  "license": "MIT",
  "keywords": ["lucid-agents", "x402", "payments"],
  "skills": "./skills"
}

Authoring best practices

• Prefer modern imports: @lucid-agents/core, @lucid-agents/http.
• Use Zod v4 only.
• Avoid outdated patterns such as agent.listen() and monolithic imports.
• Add dynamic context where needed using !`command`.
• Use $ARGUMENTS and $0, $1 for slash-command arguments.
• Keep workflows implementation-focused and testable.

Quality bar

Bad:

• "Run npm install to install dependencies."
• 50 lines teaching fetch() basics.
• Mock-only examples disconnected from real endpoints.

Good:

• "Requires: Zod v4, @lucid-agents/payments."
• Endpoint tables with verified inputs/outputs.
• Live-data examples with clear constraints and assumptions.

PR checklist

• SKILL.md exists with name and description frontmatter.
• description is third-person and includes Use when ....
• SKILL.md is under 500 lines.
• Large reference files include ## Contents.
• .claude-plugin/plugin.json has all required fields.
• author is an object with name and email.
• Uses Zod v4 and modern Lucid SDK imports.
• No bloat, no repetitive basics, no hardcoded mock flows.
• Skill clearly improves paid-agent creation outcomes.

Troubleshooting

Problem	Likely cause	Fix
Skill does not trigger	Weak description terms	Add concrete keywords and explicit Use when
YAML parse error in frontmatter	Unquoted : or invalid YAML	Quote description string and validate delimiters
Skill triggers too often	Description too broad	Narrow trigger phrasing and context
Agent misses references	Nested or unlinked docs	Link every reference directly from SKILL.md

Install

/plugin marketplace add daydreamsai/skills-market
/plugin install <skill-name>@daydreams-skills