Loreto Skills Generator
Feed any YouTube video, article, PDF, or image into the Loreto API and receive production-ready skill packages, complete with SKILL.md, test scripts, and reference stubs.
loreto-mcp
Turn any YouTube video, article, PDF, or image into a reusable Claude Code skill — without leaving your editor.
What it does
Loreto analyzes a content source and extracts structured skill packages that Claude Code can apply to future tasks. Each skill contains:
SKILL.md— Principles, failure modes, implementation steps, and architectural patternsREADME.md— Overview and usage context- Reference files — Supporting patterns and data structures
- Test script — Runnable validation for the skill's core concepts
Save skills to .claude/skills/ and Claude picks them up automatically on relevant tasks — reducing hallucinations, token usage, and re-explaining the same concepts over and over.
Sample skills
The sample-skills/ folder contains real skills generated by Loreto from a single technical video on hybrid AI architecture. Browse them to see exactly what you get:
| Skill | What it teaches |
|---|---|
designing-hybrid-context-layers | Architect hybrid retrieval systems that combine vector search, graph traversal, and structured data |
temporal-reasoning-sleuth | Enable agents to trace decision chains and reconstruct causal sequences across long time horizons |
synthesizing-institutional-knowledge | Capture and query organizational knowledge in a way AI agents can reliably reason over |
diagnosing-rag-failure-modes | Identify and fix the 7 most common RAG pipeline failure patterns |
routing-work-across-ai-harnesses | Dynamically route tasks to the right AI harness based on task type and context |
evaluating-ai-harness-dimensions | Score and compare AI harness options across speed, cost, accuracy, and controllability |
detecting-harness-lockin | Spot vendor lock-in signals early and design for portability |
benchmarking-ai-agents-beyond-models | Measure agent performance at the system level, not just model level |
auditing-intelligence-context-fit | Audit whether the right intelligence layer is solving the right problem |
Each folder includes the full SKILL.md, README.md, and any reference files — ready to drop into .claude/skills/.
Setup
1. Get an API key
Sign up at loreto.io to get your free API key (lor_...).
2. Install
pip install loreto-mcp
Or run directly without installing (requires uv):
uvx loreto-mcp
3. Configure Claude Code
User-scoped (works across all your projects) — add to ~/.claude/mcp.json:
{
"mcpServers": {
"loreto": {
"command": "uvx",
"args": ["loreto-mcp"],
"env": {
"LORETO_API_KEY": "lor_..."
}
}
}
}
Project-scoped (shared with your team) — add to .mcp.json at your project root:
{
"mcpServers": {
"loreto": {
"command": "uvx",
"args": ["loreto-mcp"],
"env": {
"LORETO_API_KEY": "${LORETO_API_KEY}"
}
}
}
}
4. Verify
Restart Claude Code and run /mcp — you should see loreto listed with generate_skills and get_quota.
Usage
Once connected, just ask Claude Code naturally:
Use Loreto to extract skills from https://www.youtube.com/watch?v=JYcidOS9ozU
Extract skills from this article and save them to .claude/skills/
Check my Loreto quota before we start.
Claude calls generate_skills, receives the full skill package, and can write the files directly to your project.
Available tools
| Tool | Description |
|---|---|
generate_skills | Extract ranked skill packages from a URL. Returns full file contents ready to save. |
get_quota | Check calls used, monthly limit, and plan for your API key. |
generate_skills parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
source | str | required | URL to analyze — YouTube, article, public PDF, or image |
source_type | str | "auto" | "auto" | "youtube" | "article" | "pdf" | "image" |
test_language | str | "python" | "python" | "typescript" | "javascript" |
include_visuals | bool | true | Embed Mermaid diagrams in SKILL.md |
context | str | null | 1–3 sentence hint to guide extraction (max 500 chars) |
themes_to_process | list[str] | null | Follow-up call: skill names from a previous response's queued themes |
Supported sources
| Source | Notes |
|---|---|
| YouTube videos | Up to 60 minutes |
| Web articles | Any publicly accessible URL |
| PDFs | Up to 100 pages |
| Images | Diagrams, whiteboards, slides (up to 20 MB) |
Configuration
| Environment variable | Required | Default | Description |
|---|---|---|---|
LORETO_API_KEY | Yes | — | Your Loreto API key (lor_...) |
LORETO_BASE_URL | No | https://api.loreto.io | Override for local development |
Plans
Free and paid plans available. See loreto.io/pricing for current limits.
License
MIT
관련 서버
SuperLocalMemory V2
Universal, local-first persistent memory for AI assistants. SQLite-based knowledge graph with zero cloud dependencies. Works with 17+ tools (Claude, Cursor, Windsurf, VS Code, etc.). 100% free forever.
llmconveyors-mcp
39 tools for the LLM Conveyors AI agent platform. Run Job Hunter, B2B Sales, ATS scoring, resume rendering, and more from any MCP client.
Overleaf
Access and analyze Overleaf projects and LaTeX files through Git integration.
YOURLS-MCP
Integrates the YOURLS URL shortening service with Claude Desktop.
Fathom AI
Access fathom ai api endpoints via mcp server and llm.
TaskWarrior MCP Server
An MCP server for managing tasks with the command-line tool TaskWarrior.
n8n MCP Server
An MCP server for interacting with n8n workflows via natural language.
calculator-mcp-server
Math evaluation, simplification, derivatives
Claude Desktop Extension
An MCP extension for the Claude Desktop application that enables automation and integration.
Pulsetic MCP Server
The Pulsetic MCP Server connects Pulsetic monitoring with AI agents and MCP-compatible tools, enabling direct access to uptime data, cron monitoring results, incident management workflows, and status page information through the Model Context Protocol (MCP). It allows teams to securely expose operational monitoring data in a structured format, making it easy to build AI-driven automation, monitoring assistants, and intelligent operational workflows without custom middleware.