Guides
Import Skills
Bring existing Skills into a workspace from a connected AI tool, a file upload, a public GitHub or GitLab repository, or an npm package, with a trust review on every import.
- Audience
- Anyone moving Skills they already have into a Fygment workspace.
- Updated
- 2026-11-01
Import turns a Skill you already have into a Fygment Skill. Each import is validated and trust reviewed. A Skill that passes is ready for you to use right away. Sharing it more widely is a separate step.
Where you can import from
Every plan can upload a Skill as a .zip file, a .skill file, or a single SKILL.md file.
Every plan can also bring in a Skill from a connected AI tool. The AI shows you what it would send and asks before it imports anything. On Solo, this is limited to 20 attempts a day per workspace. An attempt counts once you confirm the import, even if the import is then refused.
On the Team plan you can also import from a public GitHub repository, a public gitlab.com repository, or a public npm package. Support for ZIP URLs and marketplaces is planned and shows as disabled on the import page.
Imports only read public sources. For a private repository or a private package registry, download the Skill and upload the file instead.
Import from a repository or package
- 1
Preview the source
Paste the repository URL or npm package name. Fygment lists every Skill it finds, with its validation result, file count, risks, and trust review findings.
- 2
Choose the Skills to import
The first Skill found is selected for you. You can select up to three and import them together.
- 3
Fix invalid Skills with AI-assisted normalization
On the Team plan, tick the normalization box to let Fygment repair a Skill that does not pass validation. The repaired Skill still goes through the trust review.
- 4
Import
Each Skill that passes the trust review is saved as a new Skill. Fygment records its checksums, which you can see later on the Skill’s Package tab.
Who can use an imported Skill
An import that passes the trust review is approved and ready to use at once. There is no separate approval step for your own use.
A new imported Skill starts as personal: the Skill is yours. Workspace owners and admins can also view it.
If you are a member, send a request to share it with a team or the whole workspace. The request goes to the Approval Queue, and you can follow it under My requests.
If you are a workspace owner or admin, you can publish or share the Skill directly.
A Skill sent as text from an AI tool always stays personal, even for owners and admins. If a shared Skill already uses its name, the import is refused.
If the trust review recommends against the Skill, or cannot finish, an import from the import page is refused and nothing is saved. A Skill the review rates Caution, or one with a file it could not read, is saved as pending and waits for a person to say yes. See Trust review for details.
Privacy
What leaves your workspace
A full trust review sends the package contents to Azure OpenAI. Fygment keeps only a short summary of the result.
Error reports and support reports leave out package contents, prompts, tokens, credentials, and local file paths.
Related docs
- Trust review
Every Skill you import is checked by the SkillSpector trust scanner before Fygment saves it. You see one of three results: Safe, Caution or Do not install. A running scan is waited for; a scan that fails after retries imports nothing. The scanner lowers risk, but it cannot guarantee that a Skill is safe.
- Create a Skill
Creating a Skill gives you a valid package from the start, so you can focus on the instructions. The first version is approved and current as soon as you create it.
- Review and approve Skills
Members can use the Skills they import right away. Review is for sharing more widely and for new versions. Each decision is made by someone with authority over that Skill or team, and every version approval re-runs the trust review.
- Install Skills in Codex
Codex is a writable install target. Use the guided CLI path so Fygment can preview config changes and track install state.