Publish Once.
Verify Anywhere.
The trustless specification for agent skills.
Guaranteed by cryptographic consensus, not platform promises.
Get Access
auto-identitySign in with email or connect a wallet. You need credits to publish—your identity establishes verified ownership.
Publish a SkillAuthor and Document
The 4-step publish wizard walks you through identity, author info, license, documentation, tags, and category. Preview everything before publishing.
Tip: Use Markdown for your skill documentation. It renders directly on the skill detail page.
Review and Publish
Get a credit quote based on actual HBAR fees, then publish. Your skill files are inscribed via HCS-1, a version topic is created via HCS-2, and the skill is registered in the global directory.
After publishing: Request verification (2,000 or 10,000 credits) for a trusted badge after admin review.
Ready to publish? Start the publish wizard
Capabilities become on-chain standards.
A tamper-proof pipeline from your terminal to the consensus layer.
Hashing skill package...
What trustless verification enables.
When skills are published on-chain, new capabilities become possible for every participant in the ecosystem.
Verifiable Provenance
Every skill has a cryptographic origin story — who published it, when, and what changed between versions. Anyone can audit the full history.
Composable Discovery
Any agent, on any platform, can discover and verify skills through a single open protocol. No vendor lock-in, no proprietary APIs.
Tamper-Evident History
Version history is append-only. If something changes, everyone can see it. No silent updates, no surprises downstream.
Open Standard
Skills are not tied to a specific platform, vendor, or runtime. HCS-26 is an open specification that anyone can implement and extend.
Built on open standards.
HCS-26
Protocol LayerDefines the metadata schema, publishing rules, ownership verification, and discovery protocol.
HCS-2
Registry LayerProvides topic-based message ordering for immutable version histories and global discovery.
HCS-1
Storage LayerHandles content-addressed file inscription and SHA-256 integrity proofs on the consensus ledger.
Security you can verify, not just trust.
Every guarantee is enforced by the protocol itself and independently verifiable by anyone.
No Force-Pushes
Published skills are inscribed on the consensus layer. Nobody can overwrite, delete, or silently replace a version after publication.
Verify Before You Run
Every skill ships with a SHA-256 checksum. Agents and humans can independently verify that the code they downloaded matches what was published.
Cryptographic Ownership
Only the authorized identity that created a skill can publish new versions. No shared credentials to steal.
Ship skills on a trustless foundation.
Publish your first trustless skill or browse verified skills from the community.