Core capabilities
Skills
Reusable capability packages that give agents specialized instructions, workflows, and tool context.
What it is
Ottermind Skills are reusable capability packages that provide Agents with the instructions, workflows, and tool context needed for specific tasks. Think of a Skill as an operating manual that helps an Agent complete a certain type of task more reliably. Loading a Skill helps the Agent work more stably and with more control in the relevant scenario.
With Skills, you can
- Specialize agents with task-specific capabilities.
- Reuse stable workflows and reduce repeated instructions.
- Control which agents can use which capabilities through bindings.
- Manage source, version, status, and risk.
Skills are suitable for repeatable tasks with clear workflows and stable output requirements. They are not suitable when the source is unknown, permissions are unclear, or sensitive data must be handled but cannot be reviewed.
Manage your skills
In the left sidebar of Ottermind, select Skills to view installed Skills. You can search, enable, disable, or delete Skills, and view their source, version, and update time.
| Status | Meaning |
|---|---|
| Enabled | The Skill can be used by bound Agents. |
| Disabled | The Skill remains in the list but will not take effect. |
| Deleted | The Skill is removed from the current workspace. |
| Try it out | Coming soon — for quick Skill testing. |
Add new skills
On the Ottermind Skills page, click + Add skill to add a new Skill. There are three ways to add one:
In Explore, search, filter, or sort Skills before installing. Downloads, installs, and stars are only references — they do not prove that a Skill is safe. Before installing, check the source, description, version, and update time.
Use this when you already have a local Skill file. Ottermind supports ZIP archives and UTF-8 encoded SKILL.md files. SKILL.md must include YAML frontmatter with at least name and description.
Use this when the Skill is hosted at an external URL. The URL should point directly to an accessible SKILL.md or ZIP file. Do not use a regular webpage URL.
View details
In Details, review the Skill's capability scope, input requirements, output type, and whether it depends on scripts, file access, or external services.
Bind an agent
In Bindings, view which Agents can use the Skill. After binding, the Agent will automatically decide whether to use the Skill while handling a task. If the Skill is not bound to the current Agent, that Agent cannot use the corresponding capability.
Use your skills
After a Skill is installed and bound, you do not need to manually select or call it. Give the task directly to the corresponding Agent — the Agent will decide whether it needs to use any bound Skill based on the task.
Usage notes
Bindings define availability
A Skill does not automatically apply to all Agents after installation. Only Agents bound to the Skill can use it. If the result is not as expected, first check whether the Skill is enabled and bound to the current Agent.
Review third-party skills first
Third-party Skills may include scripts, external API calls, or file access. Before use, check the source, permissions, and data scope. Do not give passwords, keys, customer data, contracts, financial data, or personal identity information to an unreviewed Skill.
FAQ
Why can't my Agent use an installed Skill?
The Skill may be disabled, not bound to the current Agent, the input may not meet requirements, or the task may not match the Skill's applicable scenario.
Can I use a community Skill directly?
Not recommended. First check its source, description, permissions, scripts, and data access scope.
What should I do if import fails?
Check the file format, encoding, name, description, and whether the URL points directly to an accessible Skill file.