Stop open sourcing
your income.
Sell skills, MCP servers, CSS art, prompts, starter kits, Cowork configs — we provide the verification and paid layer for anything you ship. Mint Verified on every listing. You keep 80%.
Skills. CSS art. MCPs.
Prompts. Starter kits.
Select any listing to see it run live →
Live run — Legal Compliance Checker
Two sides of
the same bad deal.
Devs build it. Buyers ship it. Nobody gets protected.
Devs
give it away.
Revenue
per GitHub project
Distribution
strategy
Reuse rate
same skill built by
Payment layer
included
Buyers
get burned.
Malicious skills
ClawHub — shipped live
CVEs
MCP servers — 60 days
Secrets exposed
in public configs
Exec permissions
skills run with
Devs get paid. Buyers get protection. The commercial layer the free stack was never going to build — payments and trust in a single verified marketplace.
Mint
Verified.
Your agent never runs malicious code. Every skill earns its mark before it lists — or it doesn't list.
Every skill. Four gates. No exceptions.
Static Analysis
◌ READYEvery SKILL.md and associated code scanned for outbound network calls, environment variable harvesting, credential reads, and eval/exec invocations. Anything that looks like exfiltration fails the gate.
Your work
has a price.
List once. Get paid every time someone installs it.
GitHub is the world's largest free code warehouse. Developers, CSS artists, prompt engineers, and workflow builders all give their work away. MintSkills is the paid layer that changes that — for every category, not just skills.
That's £12,384 per year.
Start selling your work →80% to you. Always.
We take 20%. You keep 80%. Most marketplaces take 30–50%. That gap is not a rounding error — on £1,000 of monthly sales it is the difference between £500 and £800 in your pocket. No tiers, no performance clauses, no renegotiation when you get popular. That is the deal.
Set your own price.
List from £10 to £500 and beyond. Your skill, your rate. We do not cap earnings or force race-to-the-bottom pricing.
Your source code stays yours.
Buyers receive secure API delivery or a licence-gated install. They never see the raw source. We inject a watermark at distribution so any leak is traceable.
List once, sell forever.
One listing. Skills, CSS art, MCP servers, prompts, Cowork configs — all sold through a single verified marketplace. You build it; we distribute it.
Install. Compose. Ship.
Know what you ran.
Every skill on this marketplace has been verified before your agent touches it.
Install — early access
A single install command. Works across Claude Code, Cursor, Windsurf, Codex, and 14+ other runtimes. Exact syntax ships with your early access invite.
mintskills install <skill>coming soonSyntax confirmed before early access opens. No phantom commands.
One command.
Works in Claude Code, Codex, Cursor, Windsurf, VS Code Copilot, Gemini, Cowork, OpenClaw, and fourteen more. Install once; the runtime figures out the rest.
Every skill verified.
Your agent only ever runs code that cleared four security gates. No Mint Verified mark — no listing. Static analysis, dependency verification, content hashing, similarity scan: all mandatory, all applied before you see the skill.
Tamper-evident receipts.
Know exactly what your agent ran. Every install writes a signed receipt with the SHA-256 of what you deployed. If the skill changes post-purchase, you know immediately. If it was tampered with, we can prove it.
● All runtimes below are confirmed Mint Verified compatible.
Honest list only. We add a runtime when we have confirmed it, not when we want the logo.
Everything you'd ask
before trusting us with your code.
Honest answers. No marketing copy.
Three things, deliberately layered. First, in the north of England "mint" is everyday slang for brilliant, outstanding, excellent — so mint skills are simply great skills. Second, "minted" is a colloquial British expression for becoming wealthy — which is exactly what we want devs to do by listing here. Third, there is the Royal Mint connection: something in mint condition is verified, pristine, and trusted. That third meaning is the one the product is built on. Every listing that earns the Mint Verified mark has passed our four-stage security pipeline — it is, in every sense of the word, mint.
Four automated gates, in order. Static analysis scans every SKILL.md and associated code for outbound network calls, environment variable harvesting, credential reads, and eval/exec invocations. Dependency verification cross-references every referenced npm and pip package against canonical registries and known vulnerability databases. Content hashing stores a SHA-256 hash of the listing at publication time. Similarity analysis compares the submission against existing listings to catch near-identical uploads, reposted work, and subtle clone attacks. Pass all four or you do not list. There is no appeal process for failed gates — fix the issue and resubmit.
No. Buyers receive either a secure API delivery or a licence-gated install. The raw source is never exposed. We inject a watermark at distribution so if a leak occurs, we can trace it to the specific transaction. Your intellectual property stays yours.
We take 20%. You keep 80%. That is the deal for every listing, every sale, at every price point — from £10 to £500 and beyond. No tiers, no performance thresholds, no renegotiation when you get popular. Devs in the founding cohort lock in 80% permanently. Payouts are processed on a rolling basis. Full payment terms are confirmed at early access onboarding.
Yes, with one condition: you must own the copyright or have explicit licence rights to commercialise it. You cannot list code you do not have the right to sell, regardless of whether it is publicly available on GitHub. We verify this during the submission review. If you open-sourced your own work, you can sell a commercial licence for it alongside the free version — many successful App Store developers do exactly this.
You receive a report specifying exactly which gate failed and why. Fix the issue and resubmit — there is no penalty for a failed first submission. Repeated submissions of knowingly malicious code result in permanent account termination.
A Skill is a SKILL.md file — a structured capability package that teaches an AI agent how to perform a specific task. An MCP Server is a Model Context Protocol server that exposes tools and data sources to any MCP-compatible runtime. A Cowork config is a pre-built agent configuration for the Cowork desktop tool. All three are sold through the same verified marketplace with the same 80% revenue share and the same Mint Verified pipeline. The category affects how buyers install and use the product, not how you list or get paid.
You can delist at any time. Buyers who have already purchased retain access to the version they bought — we honour the signed receipt they received at install. You stop receiving new sales immediately upon delisting. Revenue already earned is paid out on the next payment cycle.
Every install issues a cryptographically signed licence receipt containing the SHA-256 hash of exactly what you installed, a timestamp, your licence terms, and the listing version number. If the skill you installed is ever modified without your knowledge, the hash will not match. If a dispute arises about what you ran, the receipt is the proof. This is the same principle used in software supply chain security — applied to agent skills for the first time.
Yes — and this is exactly why verification matters. Agent skills execute with the permissions of the agent runtime they run inside. That means a skill installed into Claude Code inherits Claude Code's access to your filesystem, your environment variables, and your API keys. This is true of every skill from every source — free or paid. The difference is that every Mint Verified skill has been scanned specifically for code that tries to exploit those permissions maliciously. The ClawHub incident in January 2026 — 335 malicious skills stealing exchange keys, SSH credentials, and browser passwords — happened because there was no verification gate. Mint Verified is that gate.
Your signed receipt proves what version you installed. If the listing is updated in a way that breaks functionality, you can raise a dispute and we will mediate. If the issue is with the skill itself rather than a platform update, the dev is responsible for maintaining compatibility. Refund eligibility and dispute resolution terms are confirmed at early access onboarding.
Refund policy is confirmed at early access onboarding. Generally: if a skill does not function as described and the dev cannot resolve the issue within a reasonable timeframe, you are entitled to a refund. Because skills are digital goods that can be inspected before purchase via the live demo in the catalogue, we do not offer refunds on the basis of "changed my mind."
The SHA-256 hash in your signed receipt matches the hash stored at verification time. If anyone — including the dev — modifies the skill after verification, the hashes will not match. You can verify this yourself. This is not a claim we make. It is a mathematical proof.
You can. Free skills work. The question is whether you know what you are running. A free skill from GitHub has no verification, no receipt, no accountability, and no recourse if it harvests your credentials. For personal projects that risk nothing sensitive, free is fine. For anything running against production credentials, client data, or business-critical systems — the question is not whether verified skills are worth paying for. It is whether the risk of unverified skills is worth avoiding.
No. MintSkills is an independent product built by Murmur Labs Ltd, registered in England and Wales. We are not affiliated with, endorsed by, or funded by Anthropic, OpenAI, or any AI model provider. We work across all runtimes precisely because we are independent of all of them.
No. Nothing about MintSkills involves cryptocurrency, blockchain, tokens, NFTs, or Web3. The word "mint" refers to mint condition — verified, pristine, trustworthy — and to the northern English slang for something outstanding. Our signing and hashing uses standard cryptographic primitives (SHA-256) that have been used in software security for decades. You pay with normal currency. You receive a normal licence. There is no token.
Three things, deliberately layered. First, in the north of England "mint" is everyday slang for brilliant, outstanding, excellent — so mint skills are simply great skills. Second, "minted" is a colloquial British expression for becoming wealthy — which is exactly what we want devs to do by listing here. Third, there is the Royal Mint connection: something in mint condition is verified, pristine, and trusted. That third meaning is the one the product is built on. Every listing that earns the Mint Verified mark has passed our four-stage security pipeline — it is, in every sense of the word, mint.
Four automated gates, in order. Static analysis scans every SKILL.md and associated code for outbound network calls, environment variable harvesting, credential reads, and eval/exec invocations. Dependency verification cross-references every referenced npm and pip package against canonical registries and known vulnerability databases. Content hashing stores a SHA-256 hash of the listing at publication time. Similarity analysis compares the submission against existing listings to catch near-identical uploads, reposted work, and subtle clone attacks. Pass all four or you do not list. There is no appeal process for failed gates — fix the issue and resubmit.
No. Buyers receive either a secure API delivery or a licence-gated install. The raw source is never exposed. We inject a watermark at distribution so if a leak occurs, we can trace it to the specific transaction. Your intellectual property stays yours.
We take 20%. You keep 80%. That is the deal for every listing, every sale, at every price point — from £10 to £500 and beyond. No tiers, no performance thresholds, no renegotiation when you get popular. Devs in the founding cohort lock in 80% permanently. Payouts are processed on a rolling basis. Full payment terms are confirmed at early access onboarding.
Yes, with one condition: you must own the copyright or have explicit licence rights to commercialise it. You cannot list code you do not have the right to sell, regardless of whether it is publicly available on GitHub. We verify this during the submission review. If you open-sourced your own work, you can sell a commercial licence for it alongside the free version — many successful App Store developers do exactly this.
You receive a report specifying exactly which gate failed and why. Fix the issue and resubmit — there is no penalty for a failed first submission. Repeated submissions of knowingly malicious code result in permanent account termination.
A Skill is a SKILL.md file — a structured capability package that teaches an AI agent how to perform a specific task. An MCP Server is a Model Context Protocol server that exposes tools and data sources to any MCP-compatible runtime. A Cowork config is a pre-built agent configuration for the Cowork desktop tool. All three are sold through the same verified marketplace with the same 80% revenue share and the same Mint Verified pipeline. The category affects how buyers install and use the product, not how you list or get paid.
You can delist at any time. Buyers who have already purchased retain access to the version they bought — we honour the signed receipt they received at install. You stop receiving new sales immediately upon delisting. Revenue already earned is paid out on the next payment cycle.
Every install issues a cryptographically signed licence receipt containing the SHA-256 hash of exactly what you installed, a timestamp, your licence terms, and the listing version number. If the skill you installed is ever modified without your knowledge, the hash will not match. If a dispute arises about what you ran, the receipt is the proof. This is the same principle used in software supply chain security — applied to agent skills for the first time.
Yes — and this is exactly why verification matters. Agent skills execute with the permissions of the agent runtime they run inside. That means a skill installed into Claude Code inherits Claude Code's access to your filesystem, your environment variables, and your API keys. This is true of every skill from every source — free or paid. The difference is that every Mint Verified skill has been scanned specifically for code that tries to exploit those permissions maliciously. The ClawHub incident in January 2026 — 335 malicious skills stealing exchange keys, SSH credentials, and browser passwords — happened because there was no verification gate. Mint Verified is that gate.
Your signed receipt proves what version you installed. If the listing is updated in a way that breaks functionality, you can raise a dispute and we will mediate. If the issue is with the skill itself rather than a platform update, the dev is responsible for maintaining compatibility. Refund eligibility and dispute resolution terms are confirmed at early access onboarding.
Refund policy is confirmed at early access onboarding. Generally: if a skill does not function as described and the dev cannot resolve the issue within a reasonable timeframe, you are entitled to a refund. Because skills are digital goods that can be inspected before purchase via the live demo in the catalogue, we do not offer refunds on the basis of "changed my mind."
The SHA-256 hash in your signed receipt matches the hash stored at verification time. If anyone — including the dev — modifies the skill after verification, the hashes will not match. You can verify this yourself. This is not a claim we make. It is a mathematical proof.
You can. Free skills work. The question is whether you know what you are running. A free skill from GitHub has no verification, no receipt, no accountability, and no recourse if it harvests your credentials. For personal projects that risk nothing sensitive, free is fine. For anything running against production credentials, client data, or business-critical systems — the question is not whether verified skills are worth paying for. It is whether the risk of unverified skills is worth avoiding.
No. MintSkills is an independent product built by Murmur Labs Ltd, registered in England and Wales. We are not affiliated with, endorsed by, or funded by Anthropic, OpenAI, or any AI model provider. We work across all runtimes precisely because we are independent of all of them.
No. Nothing about MintSkills involves cryptocurrency, blockchain, tokens, NFTs, or Web3. The word "mint" refers to mint condition — verified, pristine, trustworthy — and to the northern English slang for something outstanding. Our signing and hashing uses standard cryptographic primitives (SHA-256) that have been used in software security for decades. You pay with normal currency. You receive a normal licence. There is no token.
Still have a question?
Get in early.
Before the market sets.
MintSkills is in private beta. Devs in the first cohort lock in 80% revenue share permanently — that rate is guaranteed for life for founding members. Tell us what you are here for and we will place you in the right cohort when access opens. One email. No noise.