# Skills (/cli/skills)





Mogplex ships a set of [Anthropic Agent Skills](https://docs.claude.com/en/docs/agents-and-tools/agent-skills) that teach a skill-aware agent how to help users drive the Mogplex CLI.

<Callout>
  The current CLI is interactive-only — there is no headless `exec` surface
  for an agent to call. These skills are **guidance skills**: they teach an
  agent to recommend the right slash command, walk a user through the login
  screen, and help author repo-local config (`AGENTS.md`,
  `permissions.json`). Agents do not drive the cockpit directly.
</Callout>

## Source of truth [#source-of-truth]

The canonical skill files live at `skills/<name>/SKILL.md` in [`webrenew/mogplex-docs`](https://github.com/webrenew/mogplex-docs). The pages under this section mirror those files.

## Skills [#skills]

<Cards>
  <Card title="using-mogplex-cli" href="/cli/skills/using-mogplex-cli" description="Umbrella skill. Routes the agent to the right guidance when the user asks for a Mogplex CLI action." />

  <Card title="mogplex-slash" href="/cli/skills/mogplex-slash" description="Tell the user which slash command to type in the cockpit composer." />

  <Card title="mogplex-auth" href="/cli/skills/mogplex-auth" description="Walk the user through the in-app login screen, env-var escape hatches, and credential troubleshooting." />
</Cards>

## Install [#install]

The skill files are hosted at [`webrenew/mogplex-docs`](https://github.com/webrenew/mogplex-docs) under `skills/`. Pick one fetch method, then copy into your agent host.

### Claude Code [#claude-code]

```bash
# Option A — full clone
git clone https://github.com/webrenew/mogplex-docs.git
SKILLS=mogplex-docs/skills

# Option B — sparse download of just the skills/ tree
npx -y degit webrenew/mogplex-docs/skills mogplex-skills
SKILLS=mogplex-skills
```

Then:

```bash
# Global
mkdir -p ~/.claude/skills && cp -R "$SKILLS"/* ~/.claude/skills/

# Or per-project
mkdir -p .claude/skills && cp -R "$SKILLS"/* .claude/skills/
```

### Claude Agent SDK [#claude-agent-sdk]

Skills picked up automatically via `settingSources: ["user", "project"]`:

```ts
import { query } from "@anthropic-ai/claude-agent-sdk";

const response = query({
  prompt: "How do I switch models in Mogplex?",
  options: { settingSources: ["user", "project"] },
});
```

### Cursor / opencode / other hosts [#cursor--opencode--other-hosts]

Copy the body of each `SKILL.md` into the host's project-rules or system-prompt-include mechanism. The frontmatter is Claude Code / Agent SDK-specific; the markdown body is portable.

## Design principles [#design-principles]

* **Guidance, not execution.** Skills tell the user what to do in the cockpit. They do not try to drive the TUI.
* **Auth handoff.** Auth lives in the cockpit's login screen. Skills recommend `/login` and explain env-var precedence.
* **Repo-local config is fair game.** Skills can help the user author `AGENTS.md` and `~/.mogplex/projects/<repo-slug>/permissions.json` — those are plain files.
* **No silent writes.** Anything that mutates user-owned state is recommended to the user, not done by the agent.