Introduction
What SenClaw is, what it publishes, and what it does and does not promise.
SenClaw is a registry for the things an AI coding agent installs: skills, plugins, workflows and apps. The store lists them; SenClaw stores the bytes and the metadata that make an install checkable.
The difference between installing from here and cloning a repository is that every package has an immutable version, a server-computed checksum, a signature, and a declaration of what it expects to touch. None of those exist when you git clone and point an agent at the result.
What you can publish
| Kind | What it is |
|---|---|
skill | A markdown instruction package the agent loads when it matches the task. |
app | A desktop or CLI binary, published across a platform matrix with auto-update. |
plugin | Skills, commands, subagents, hooks and MCP servers bundled and installed as one unit. |
workflow | A JavaScript script that orchestrates subagents deterministically. |
See Kinds and platforms for where each one installs.
What SenClaw promises
- Immutable versions. A published
name@versionalways means the
same bytes. It can be deprecated or yanked, never replaced.
- Checksums that were computed, not claimed. The SHA-512 is produced by
streaming the stored object server-side, not copied from what the publisher said.
- Signatures. The registry signs each artifact; the client verifies against a
pinned key before writing to disk.
- Declared permissions with a diff on upgrade. See Permissions.
- No install scripts. Nothing executes at install time. There is no flag to
turn this off, because there is nothing to turn off.
What it does not promise
That a package is safe. Automated checks find known patterns, not intent. A listing is not an endorsement, and the verified checkmark means SenClaw confirmed who a publisher is — nothing at all about their code.
If you find something that should not be here, use the Report button on its package page, or contact us. See Reporting and moderation.