Buyers resell. You don’t see a cent.
An unprotected agent ZIP can be forwarded, resold, or uploaded anywhere. Single-use licenses make your work non-transferable — cryptographically.
Now in public beta
CLAUDE SKILLS PROTECTION
Stop losing revenue when your Claude skills get copied
AgentVerif gives your buyers instant proof the skill is authentic, untampered, and safe to run.
pip install agentverif-sign
First 100 Claude skills creators get free access
Worth €100/year — yours at zero cost forever. No card required.
Zero risk. Try it on your next skill. If it doesn’t help you keep more revenue, you lose nothing.
117
agents certified
0
unsafe agents signed
OWASP
LLM Top 10 compliance
Why Claude skills creators switch to AgentVerif
You spent hours building a skill that clients actually pay for. Then someone copies it, tweaks it, or shares it for free — and your income disappears.
Buyers keep asking: “Is this skill safe and original?”
One verification badge ends the questions and builds instant trust.
One verification badge ends the questions and builds instant trust.
The moment you deliver the skill, you lose control
Tamper detection + MCP auto-verify blocks modified versions before they run.
Tamper detection + MCP auto-verify blocks modified versions before they run.
Copies kill your recurring revenue
Verified skills convert better, reduce refunds, and protect the income you already earn.
Verified skills convert better, reduce refunds, and protect the income you already earn.
Your agent is one button away from being stolen.
Unlimited copies from one purchase.
Without activation limits, one buyer becomes a thousand. Pro licenses let you set exactly how many times your agent can be activated.
No proof. No recourse.
If your agent gets stolen and redistributed, you have nothing. A revocable agentverif license gives you proof of issue, proof of abuse, and the ability to kill the license instantly.
It takes 30 seconds
One command. A license ID. Verification everywhere buyers run your skill.
Run one command to sign your skill
CLI or GitHub Action. Your ZIP is scanned against the OWASP LLM Top 10, then signed in seconds.
agentverif-sign signGet a clean badge + License ID
A unique License ID is generated, a cryptographic
hash registered, and SIGNATURE.json injected into your ZIP.
Buyers see VERIFIED ✓ Safe to run inside Claude
Buyers verify via CLI, MCP, or directly inside Claude. If your skill leaks, revoke in one command — it stops everywhere instantly.
sha256 hash-lockedCLI quickstart
As a seller
agentverif-sign sign ./agent.zip
# Signed successfully
# License: AC-84F2-91AB
# Tier: indie
As a buyer
agentverif-sign verify ./agent.zip
# VERIFIED
# License: AC-84F2-91AB
# Verify: verify.agentverif.com/…
Output is JSON-compatible for CI/CD pipelines, LLM tool calls, and MCP integrations
(agentverif-sign verify --json).
The window to set the standard is open. Not for long.
The AI agent marketplace is forming now. Certified agents will win by default.
Your competitor is already protecting their skills.
Verified skills convert better and command higher prices. First-mover advantage in trust is real.
One copied skill kills recurring revenue.
A single redistribution wipes out months of income. Tamper detection proves you shipped the original.
Enterprise teams need proof before deploying.
Compliance teams require audit trails. If you can’t prove your skill is authentic and unmodified, you don’t get deployed.