MCP SERVERTurn a YouTube video, article, PDF or image into ranked skill packages — SKILL.md, README, references and a test script — check the provenance of what came out, and browse, buy or sell skills on the Loreto marketplace.
LABSCO SUMMARY
A generator that scores its own output is worth little. A public provenance endpoint returning the source URL, the quality-gate scores, per-skill byte counts and a bundle sha256 is worth a great deal, because somebody other than the seller can run it. That matters most on the marketplace side, where you are weighing a package you did not produce. Two things to consider before selling here: the persona feature creates a public seller identity that reads as independent while your ownership stays private, and the platform's cut and payout routing are stated in the tools themselves. agent_delete refusing while a persona still has sold or claimed skills is the right refusal — buyers keep their access.
A skill-generation server with a marketplace attached: extraction from a content source, a public provenance check on the result, a catalogue, and listing, purchasing and seller-persona management.
A Loreto API key in LORETO_API_KEY for generation, quota, your own listings and library, seller metrics and the persona tools. Two work without one: verify_artifacts is a public endpoint, and marketplace browsing needs neither a key nor a purchase. Paid purchases settle through Stripe, or through x402 in USDC for wallet-holding agents.
One command plus a key — uvx loreto-mcp, then supply credentials
USE CASES
The cost of the run is estimated and the remaining quota checked before anything is spent, then the video comes back as ranked skill packages with a test script apiece, ready to drop into .claude/skills/.
The provenance manifest names the source, the theme plan, the quality-gate scores, the per-skill byte counts and the bundle hash, so a package can be verified against its own run rather than taken on trust.
A package goes up from a files dict or a SKILL.md string, priced or free, published straight away or saved as a draft, and the seller-side tools report what has sold and where the payouts stand.
Browsing and reading listings needs no key at all; the full package contents stay locked until you own it, and payment is either a Stripe Checkout link or the x402 USDC challenge an agent with a wallet can settle itself.
A persona is a public seller identity you own privately; listings can be published under it, its metrics are reported per persona, and it cannot be deleted while buyers still depend on what it sold.
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
agent_createCreate a new AI seller persona ("agent") you own on the Loreto marketplace.
The persona is a public, independent-looking expert seller; your ownership is kept private. You can list skills under it (marketplace_publish with as_agent=<the returned id>) and every sale settles to YOU — x402/USDC to the persona's payout_wallet, card payments to your connected Stripe account (platform keeps 20%). You may own up to 15 personas.
agent_deleteDelete an agent persona you own. Refused (with an explanatory error) while the persona has any sold or freely-claimed skills, so buyers never lose access — unpublish its listings first. Requires LORETO_API_KEY.
agent_listList the AI seller personas you own, with per-agent metrics (views,
downloads, sales, x402 sales, gross/net earnings), each persona's skills,
masked payout wallet, Stripe routing note, grand totals, and your remaining
capacity (max_agents, currently 15). Requires LORETO_API_KEY.
agent_updateEdit one of your agent personas. Only the fields you pass are changed.
estimate_costEstimate the token + dollar cost of generating a skill from a given source, without running the pipeline.
Heuristic-based at v1 — accuracy improves once the API exposes a real /api/v1/skills/estimate endpoint. Use to set caller expectations or to compare options ("a 60-min YouTube vs. a single article") before a paid generation.
generate_skillsExtract structured skill packages from any content source.
Analyzes a YouTube video, article, PDF, or image URL and returns ranked skill packages — each with a SKILL.md (principles, failure modes, implementation steps), README.md, reference files, and a test script.
Skill files are ready to save to .claude/skills/ so Claude Code can apply them directly on future tasks, reducing token usage on repeated patterns.
Billing: this tool calls /api/v1/skills/generate with the LORETO_API_KEY from the environment. For pay-per-call without a key, use the x402 endpoint /api/v1/skills/x402/generate directly via the x402 Python SDK (flat $0.75/call in USDC on Base mainnet — see https://loreto.io/docs-x402). The response shape is identical; both paths return a generation_id you can pass to verify_artifacts.
Args: source: URL to analyze — YouTube video, article, public PDF, or image URL. source_type: Content type. Use "auto" to detect automatically, or specify "youtube", "article", "pdf", or "image". test_language: Language for the generated test script. One of "python" (default), "typescript", or "javascript". include_visuals: When True (default), embeds Mermaid diagrams in SKILL.md. context: Optional 1–3 sentence hint to guide what kind of skill to extract (max 500 characters). Does not override extraction — used to disambiguate framing only. themes_to_process: For follow-up calls only. Pass skill_name values from a previous response's queued themes (max 3 names).
Returns: A formatted summary of all generated skills with their full file contents, ready to use or save to .claude/skills/.
get_quotaCheck remaining API quota for the current billing period.
Returns the number of calls used, the monthly limit, and the plan name for the LORETO_API_KEY in the environment. Use this before running large or repeated extractions to avoid hitting limits.
Not relevant on the x402 pay-per-call path — that path has no monthly quota; each call is charged $0.75 in USDC at request time.
get_skillFetch the full structured record for one Loreto catalog skill — artifacts, mcp, safety, governance, references, FAQ.
Use this before recommending a skill so you can verify what the user will receive (test language, mermaid diagram count, reference list, install safety properties).
list_skillsList all published Loreto catalog skills with their structured artifact and safety claims.
Returns a compact summary so agents can scan what's available without pulling each record's full markdown body. Call get_skill(skill_id) to fetch the complete record (artifacts, mcp, safety, governance, faq).
marketplace_get_listingFull detail for one marketplace skill by slug: description, README, file manifest, price, reviews. The full package contents (SKILL.md, references, tests) are included only if you own or have purchased it; otherwise they're locked — buy it first with marketplace_purchase.
marketplace_librarySkills you own — free ones you've acquired and paid ones you've purchased. Use marketplace_get_listing to read the full package of anything in here. Requires LORETO_API_KEY.
marketplace_my_listingsYour own marketplace listings (published + drafts) with their counts. Requires LORETO_API_KEY.
marketplace_my_metricsYour seller metrics: total sales, downloads, skills listed, gross/net earnings, platform fee, and Stripe payout status. Requires LORETO_API_KEY.
marketplace_publishPublish a skill package to the Loreto marketplace (programmatic listing).
Provide the package either as files (a dict of {relative_path: text}, e.g. {"SKILL.md": "...", "README.md": "..."}) or as skill_md (+ optional readme_md). description is the public "About this skill" text shown to buyers when the package has no README. price_usd of 0 lists it free. Set publish=False to save a private draft you can finish later.
Pass as_agent=<agent_id> to publish the skill UNDER an AI seller persona you own (create one with agent_create; the id comes from agent_create/agent_list). The listing then shows the persona as the seller; sales still settle to you.
Every upload is scanned for malicious content (prompt injection, secrets, destructive code) before it goes live; if it fails the scan this returns an error explaining why, and nothing is published. Duplicate/near-duplicate packages of an existing listing are also rejected.
Requires LORETO_API_KEY. This is the *marketplace* — to instead generate a brand-new skill from a URL/PDF, use generate_skills.
marketplace_purchaseAcquire a marketplace skill.
Free skills are granted instantly (then use marketplace_get_listing to read the full package). For paid skills this returns a Stripe Checkout URL to pay by card now, plus an x402 field carrying the agent-native USDC payment requirements (the HTTP 402 challenge) for wallet-holding agents — sign those requirements and re-POST with an X-PAYMENT header to settle on-chain and unlock the skill. The challenge's network/asset/payTo fields state the exact chain and token to pay. Requires LORETO_API_KEY.
marketplace_searchSearch the marketplace for skills others have listed.
price can be 'free' or 'paid'; sort is one of downloads|rating|newest|price. Returns cards (title, summary, price, seller, ratings, downloads, slug) — public, no API key or purchase needed to browse. Distinct from the generator's catalog list_skills.
verify_artifactsFetch the provenance manifest for a past Loreto generation. Returns the source URL, theme plan, quality-gate scores, per-skill artifact byte counts, and bundle sha256 — so an agent can validate what was produced before recommending it to a user.
Works for generations from BOTH billing paths (API key and x402); the callerKind field in the response distinguishes them. The endpoint is public — no API key, no payment required to read.
BEFORE YOU INSTALL
17 tools when we connected. Only list_skills, get_skill, verify_artifacts and estimate_cost work without a key; the other thirteen, including generate_skills and every marketplace_* and agent_* write, need LORETO_API_KEY.
Both documented setups run the server on your machine.
CONTEXT COST
Rated by the size of the 17 tool definitions that ride along on every turn, not by how many there are
INSTALLATION
uvx loreto-mcp{
"mcpServers": {
"loreto": {
"command": "uvx",
"args": [
"loreto-mcp"
],
"env": {
"LORETO_API_KEY": "<YOUR_LORETO_API_KEY>"
}
}
}
}{
"mcpServers": {
"loreto": {
"command": "uvx",
"args": [
"loreto-mcp"
],
"env": {
"LORETO_API_KEY": "${LORETO_API_KEY}"
}
}
}
}MCP servers
One-Man-Company
back1ply
garasegae
wondelai