Import a Claude skill from a URL

Paste a Claude-compatible skill manifest URL; Platos fetches, validates, registers, and exposes it. Same flow whether the manifest is a .json file, a zip with a manifest.json at the root, or a GitHub raw URL.

The goal

An imported skill that shows up in your skills list, with its tools available for agents to enable.

Steps

Find the manifest URL.

Most Claude skills are hosted on GitHub. Use the raw URL (e.g. https://raw.githubusercontent.com/example/my-skill/main/manifest.json).
Open the import page.

Sidebar -> Skills -> "New skill" -> "Import from URL".
Paste and import.

Paste the URL. Click "Import". Platos:
- Fetches the URL with SSRF guards (private IPs are rejected).
- Parses the manifest (SkillManifestParser).
- Validates required_env against your environment's variables.
- Writes a PlatosSkill row scoped to the project.
Link required env vars.

If the skill declares required_env: ["STRIPE_API_KEY"] and that key is missing, the dialog walks you to add it via Add a provider key for that key.
Enable per agent.

Open the agent. Skills tab -> toggle the new skill. Agent now exposes the skill's tools and prompt block.

Verify

The skill row appears under /skills.
Toggling on for an agent surfaces the skill's tools in the agent's Tools tab.
A test chat turn that prompts the skill to act produces tool calls visible in Postman mode.

SSRF defence

The importer rejects URLs that resolve to private IP space:

10.0.0.0/8
172.16.0.0/12
192.168.0.0/16
169.254.0.0/16 (link-local + AWS IMDS)
127.0.0.0/8

Redirects are followed but each redirect target is re-validated. EOBD-9/10 closed the SSRF gaps in this path.

Update an imported skill

Re-importing the same URL upserts the skill if the version differs. The runtime keeps the old version alongside the new and lets agents migrate at their own pace; bump the agent's skill pin to switch.

Next steps

Quickstart if you have not enabled the skill on an agent yet.
Create your first agent and toggle the new skill on.