Skills
Reusable instruction and capability bundles an Agent can be given.
A Skill is a bundle of files — instructions, reference material, optionally executable tooling — that an Agent can be bound to, the same way it can be bound to a Knowledge Base or a Tool. Studio → Settings → Skills is where you upload, review, and manage them.
Creating a skill
Upload a ZIP archive (up to 10 MB) containing the skill's files, including a frontmatter file declaring its name and description. Elizon extracts and static-scans the archive before it can go live.
Review and activation
A new skill starts in draft status. Getting it to active requires:
- Passing the static scan — an automated check of the uploaded files. A skill that fails can't be activated until the issue is fixed and re-uploaded.
- Admin approval, if the skill requests tool access — a skill that declares
allowedTools(i.e. wants to call specific tools, not just supply instructions/reference material) needs an Admin to explicitly approve it, separate from activation itself. A skill with no tool access can activate as soon as it passes the scan.
| Status | Meaning |
|---|---|
draft | Uploaded, not yet active. |
active | Bindable to Agents. |
inactive | Deactivated — existing bindings are unaffected until you rebind. |
error | Failed the static scan or another processing step. |
Deactivating a skill shows how many Agents currently have it bound, so you know the blast radius before you do it.
AI-assisted building
Rather than hand-writing a skill's files, Studio can walk you through building one conversationally — describe what you want the skill to do, and it generates the frontmatter and content for you to review before saving.
Catalog
Beyond what your organization has built itself, a shared catalog of ready-made skills is available to install directly, without writing anything from scratch.
Download
Any skill can be downloaded as a ZIP — useful for inspecting exactly what an Agent has access to, or for editing offline and re-uploading.