QuickCreator Skill Developer

Develop and maintain skills within the QuickCreator ecosystem using the QuickCreator Skill MCP.

Prerequisites

Before using this skill, two things must be set up:

1. Developer Token

The user needs a QuickCreator developer token:

Go to QuickCreator Agent Dev
Open Settings → Create Token
Choose read, write, and publish permissions
Copy the generated token

2. Install the Skill MCP

The @quickcreator/skill-mcp package must be configured. Add to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "quickcreator-skill": {
      "command": "npx",
      "args": ["@quickcreator/skill-mcp"],
      "env": {
        "QC_API_TOKEN": "<developer_token_from_step_1>",
        "QC_API_URL": "https://api-dev.quickcreator.io/ai-blog-chat-service"
      }
    }
  }
}

The developer token is exchanged for an API token automatically by the MCP.

Temporary note (development phase):
For now we intentionally hardcode
QC_API_URL = "https://api-dev.quickcreator.io/ai-blog-chat-service"
in ~/.cursor/mcp.json because @quickcreator/skill-mcp is still in active development.
Once the MCP is officially released, remove this override and rely on the default production URL instead.

MCP Tools Reference

Tool	Description
`list_skills`	List skills by category: `personal`, `builtin`, `marketplace`, `installed`
`search_marketplace`	Search marketplace by tag (sorted by publish time or downloads)
`get_skill`	Get skill detail and full file tree
`get_skill_file`	Read a specific file from any skill
`create_skill`	Create a new empty personal skill
`fork_skill`	Fork a builtin/marketplace/installed skill to personal
`update_skill_file`	Update a file in a personal skill
`create_skill_file`	Add a new file to a personal skill
`delete_skill`	Delete a personal skill
`publish_skill`	Publish a personal skill to the marketplace
`update_published_skill`	Update an existing marketplace listing
`install_skill`	Install a marketplace skill
`uninstall_skill`	Uninstall a marketplace skill

MCP Tool Usage Rules (VERY IMPORTANT)

To avoid common MCP errors (especially "Input validation error" / missing required fields), always follow these rules when calling quickcreator-skill tools:

Always read the tool schema before first use in a session
- Use the local filesystem (like Read) to open the corresponding descriptor, e.g.
  mcps/user-quickcreator-skill/tools/list_skills.json
- Inspect:
  - required fields
  - Allowed enums
  - Whether an arguments object is mandatory
Never call CallMcpTool without an arguments object when the schema defines any properties
- Even if only one field is required, you must pass arguments:{...}.
- Bad (missing category, causes invalid_type / Required error):
```
{
  "server": "quickcreator-skill",
  "toolName": "list_skills"
}
```
- Good:
```
{
  "server": "quickcreator-skill",
  "toolName": "list_skills",
  "arguments": { "category": "marketplace" }
}
```
Always respect enums and field names from the schema
- Example: list_skills.category must be one of
  "personal" | "builtin" | "marketplace" | "installed".
- Example: fork_skill requires both:
  - skillId (e.g. "mk_XXXX", "p_XXXX", "i_XXXX", "sk_XXXX")
  - source ∈ "marketplace" | "builtin" | "installed".

Use explicit examples for each core tool

list_skills:

{
  "server": "quickcreator-skill",
  "toolName": "list_skills",
  "arguments": { "category": "personal" }
}

create_skill:

{
  "server": "quickcreator-skill",
  "toolName": "create_skill",
  "arguments": {
    "name": "ecommerce-product-image-batch",
    "description": "Generates e-commerce product images in batches from text prompts or product data."
  }
}

publish_skill:

{
  "server": "quickcreator-skill",
  "toolName": "publish_skill",
  "arguments": {
    "personalSkillId": "p_xxx",
    "authorName": "Your Name",
    "tags": ["ecommerce", "images"],
    "version": "1.0.0"
  }
}

On any MCP validation error, re-open the tool schema and fix arguments instead of retrying blindly
- Look for:
  - expected vs received types in the error
  - The property path (e.g. ["category"]) to see which field is wrong or missing.

These rules are specifically to prevent the most frequent mistakes seen when driving QuickCreator via MCP.

Skill ID Prefixes

Prefix	Type	Access
`sk_`	Built-in	Read-only
`mk_`	Marketplace	Published by users
`p_`	Personal	Editable workspace
`i_`	Installed	Read-only copies

Core Workflows

Browse & Discover

list_skills(category="marketplace")    # Browse marketplace
list_skills(category="personal")       # View your skills
search_marketplace(tag="marketing")    # Search by tag
get_skill(skillId="mk_xxx")           # View skill details

Create a New Skill

create_skill(name="my-skill-name") — creates an empty personal skill (p_ prefixed)
create_skill_file(...) — add SKILL.md with proper frontmatter
Add reference files, scripts as needed via create_skill_file
Before publishing: run the Pre-Publish Checklist
publish_skill(personalSkillId="p_xxx")

Fork & Modify

fork_skill(skillId="mk_xxx", source="marketplace") — fork to personal
get_skill(skillId="p_xxx") — inspect the forked skill
update_skill_file(...) — modify files
Publish when ready

Update an Existing Skill

get_skill(skillId="p_xxx") — read current state
update_skill_file(...) — modify files
If already published: update_published_skill(marketplaceSkillId="mk_xxx", personalSkillId="p_xxx")

Delete

delete_skill(personalSkillId="p_xxx")

Skill Development Standards

ALL skills created or updated MUST comply with these standards. Read skill-standards.md for the full specification.

Hard Rules (Quick Reference)

ALL content in English — name, description, SKILL.md body, headings, step descriptions, notes, reference files. The ONLY exception: preserving non-English text inside original prompts verbatim.
name field format — lowercase a-z, digits 0-9, hyphens - only. No uppercase, spaces, underscores, or unicode. Must not start/end with hyphen. No consecutive hyphens. Max 64 chars. Directory name must match.
description field — max 1024 chars, English, third-person. Describe WHAT and WHEN. Include trigger keywords.
No hardcoded secrets — all API keys loaded from environment variables.
Follow Agent Skills spec strictly.

Script Requirements

If a skill has a scripts/ directory, include a requirements.sh file that declares all dependencies. The sandbox is Ubuntu with standard Python pre-installed.

#!/bin/bash
# requirements.sh — runs at sandbox startup
pip install google-genai requests
apt-get update && apt-get install -y jq

Pre-Publish Checklist

Run this checklist before every publish_skill or update_published_skill call.

Publish Readiness:
- [ ] `name` field: lowercase, hyphens, digits only; no leading/trailing/consecutive hyphens; ≤64 chars
- [ ] `description` field: English, ≤1024 chars, includes WHAT + WHEN + trigger keywords
- [ ] ALL content (headings, steps, notes, references) is in English
- [ ] No hardcoded API keys or secrets (use env vars)
- [ ] SKILL.md has valid YAML frontmatter with `name` and `description`
- [ ] SKILL.md body is under 500 lines
- [ ] Reference files are one level deep (linked directly from SKILL.md)
- [ ] If scripts/ exists: requirements.sh is present and lists all dependencies
- [ ] Consistent terminology throughout
- [ ] Follows Agent Skills spec (https://agentskills.io)

If any item fails, fix it before publishing.

Available Tools for Generated Skills

Generated skills can use built-in tools within the QuickCreator platform. For the full parameter reference, see tool-reference.md.

Tool	Purpose
`nano-banana-pro-image`	Text-to-image and image-to-image generation
`openai-image`	AI image generation from text prompts
`query_image_from_knowledge_base`	Query images from user knowledge base
`query_question_from_knowledge_base`	Retrieve information from knowledge base
`query_question_from_web`	Web search and research
`ask_questions_to_user`	Structured user input collection
`shell_execute`	Execute bash shell scripts in sandbox
`code_execute`	Execute Python or JavaScript code in sandbox

Video Generation

Video generation uses Google Veo SDK via code_execute — do NOT use a generate_video tool. A reusable template is at scripts/generate_video.py. See tool-reference.md for full Veo parameter docs.

Skill File Structure

Every skill follows this layout:

skill-name/
├── SKILL.md              # Required — main instructions
├── reference.md          # Optional — detailed docs
├── examples.md           # Optional — usage examples
├── requirements.sh       # Required if scripts/ exists
└── scripts/              # Optional — utility scripts
    └── helper.py

SKILL.md Template

---
name: my-skill-name
description: Does X when the user needs Y. Use when working with Z or when the user mentions A, B, or C.
---

# My Skill Name

## Instructions
Step-by-step guidance for the agent.

## Examples
Concrete usage examples.

quickcreator-developer-skill