> For the complete documentation index, see [llms.txt](https://alludium.gitbook.io/alludium-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://alludium.gitbook.io/alludium-docs/administration/10.-skills/10.4-assigning-skills-to-agents.md).

# Assigning Skills to Agents

Skills only take effect when they are **assigned** to an agent. An agent can only see and activate skills that have been explicitly attached to it — there is no global skill pool.

***

#### Assigning Skills via the Agent Builder

Skills are managed through the **Skills** tab in the Agent Builder's Configuration panel. To access it:

1. Open **Agent Builder** from the main navigation.
2. Select an existing agent or click **+ Create Agent** to start a new one.
3. Click the **Configuration** button in the top tab bar (alongside Test, Deploy, and Share) to open the Configuration panel.
4. Within the Configuration panel, click the **Skills** tab in the sub-tab bar (alongside Details, Prompt, AI Model, Integrations, Files, and Automations).

From the Skills tab you can:

* **Browse and add skills** — The "Available skills" section lists workspace, platform, and GitHub-linked skills that can be assigned. Use the **Assign** button to attach a skill to the agent.
* **Open the full library** — Click **Open library** to browse the complete Skills Library in a new view.
* **Import skills** — Click **Import** to bring in a skill from a GitHub repository.
* **Create new skills** — Click **+ Create skill** to author a new skill directly from the agent builder.
* **Remove skills** that are no longer needed
* **Set the activation mode** for each assigned skill
* **Order skills by priority** to control which skills are preferred when multiple could apply

***

#### Activation Modes

Each assigned skill has an activation mode that controls when it loads into the agent's context:

| Mode       | Behavior                                                                                                              |
| ---------- | --------------------------------------------------------------------------------------------------------------------- |
| **Auto**   | The agent decides when to activate the skill based on conversation context. This is the default and recommended mode. |
| **Always** | The skill is loaded into context on every turn, regardless of the conversation topic.                                 |

**When to use Auto:** Most skills should use Auto mode. The agent receives a lightweight summary of all assigned skills and activates the relevant one when the conversation topic matches.

**When to use Always:** Use Always mode for skills that should apply to every interaction — such as output formatting standards, compliance guidelines, or tone-of-voice rules.

***

#### How Activation Works at Runtime

When an agent with assigned skills receives a message:

1. The agent sees a lightweight catalog of its assigned skills (name and description only — minimal token overhead).
2. If the conversation topic matches an assigned skill, the agent activates it using an internal `activate_skill` action.
3. The full skill instructions are loaded into context for that turn.
4. The agent follows the skill's methodology to complete the task.
5. Skill activation persists across conversation turns — once activated, the skill remains available for the duration of the conversation.

The agent's base system prompt is never modified by skill operations. Skills are injected as an ephemeral overlay that does not alter the agent's core configuration.

***

#### Assigning Skills via the Agent Configurator

The Agent Configurator — the conversational assistant on the left side of the Agent Builder — is also skill-aware. During a configuration conversation, the configurator can:

* **Discover** skills relevant to the agent you are building
* **Recommend** skills based on the agent's purpose and tools
* **Assign** skills to the agent on your behalf

This provides a natural-language alternative to the manual Skills tab workflow. For example, you can ask the configurator: *"Add a competitive analysis skill to this agent"* and it will search the workspace library, recommend a match, and assign it.

***

#### Skills and Deployment

Skills follow the same deployment model as other agent configuration:

* **DEV mode** — The agent uses skills as currently configured (live state).
* **PROD mode** — When you deploy an agent, its assigned skills are snapshotted at deploy time. The production agent reads from this snapshot, not from live skill state. This ensures production behavior is stable even if you edit skills later.

***

#### Best Practices

* **Start with 1-3 skills per agent** — Too many skills increase context overhead. Assign only skills that are directly relevant to the agent's purpose.
* **Use Auto mode by default** — Let the agent decide when skills are relevant rather than loading everything on every turn.
* **Order by priority** — If an agent has multiple skills that could apply to similar topics, the priority order determines which is preferred.
* **Test in DEV before deploying** — Verify that skill activation works correctly in DEV conversations before deploying to PROD.

***

**Next Steps**

Now that you understand how to create, import, and assign skills, continue to **Projects** to learn about the new workspace container for persistent context and collaboration.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://alludium.gitbook.io/alludium-docs/administration/10.-skills/10.4-assigning-skills-to-agents.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.