Skip to main content
Agents are a dedicated environment for getting a specific set of tasks done. Skills are reusable instructions that load automatically when you need them. Skills extend what agents can do: if you move generic rules like tone of voice, output formats, and brand guidelines into skills, agents become leaner and focus only on what makes them specific. In some cases, a skill can replace an agent entirely. If your team opens an agent just to get properly formatted output, that should be a skill instead. Rule of thumb: if users have to consciously “go somewhere” for something that should always just work, it belongs in a skill.
Project instructions apply only to project chats. Skills apply to normal chats and are not limited to a single project.
Skills can trigger automatically via semantic matching (not keywords). You can also manually steer by tagging a skill via the @ menu.
No. Triggering is semantic and driven primarily by the skill description, not keyword matching or the skill name. Names help humans manage skills.
Yes. Attach skills to agents in the agent configuration under Actions.
Skills active in normal chats are not automatically applied in project chats or agent chats, so explicitly adding them there matters.
Globally active skills auto-apply to all chats. You can disable individual skills for a session via the skills indicator, and you can scope skills by attaching them to agents or projects.
No formal rulebook, but typical priority is: agent prompt first, then skills, then custom instructions. The model generally follows the most specific instruction.
Yes. Max 20 active skills per user; admins can force-apply up to 8 workspace-wide skills. Too many active skills can cause wrong-skill selection or unpredictable behavior; the limit mitigates this.
Admins can decide in the role settings which roles can share and create skills. Only admins can share workspace-wide or force-apply skills to everyone.
Upload either a plain SKILL.md or a .zip skill bundle with supporting files (templates/reference docs/scripts). ZIP limit is 25 MB; uncompressed limit is 10 MB. Scripts are reference only (not auto-executed). Disallowed file types are skipped.
It must be SKILL.md.
Yes, if they’re in SKILL.md format or packaged as a .zip.
Skills are model-agnostic.
Yes. There is a “Skill Creator” skill that helps generate skills in Langdock.
Create one skill per customer, or build one skill that contains rules for selecting the right guideline set per customer.
System skills can not be deactivated.