Slash Commands (Skills & Prompts)

Typing / in a Chat Node input opens the slash menu — a quick keyboard-driven way to configure the conversation and pull in your reusable prompt library.

---

What's in the menu

Skills, Personas, and MCPs each get their own browsable section, with your reasoning presets and prompts alongside as regular entries:

1. Skills — your installed agent skills ("Browse available skills")
2. Personas — apply a persona to this chat without leaving the keyboard ("Set the persona for this chat")
3. MCPs — enable or disable your configured MCP servers for this conversation ("Toggle MCP tool servers")
4. Reasoning presets — Reasoning: Low, Medium, High, Max, for models that support extended thinking (see Models & Providers)
5. Prompts — your local library of reusable markdown prompts

Each section row shows how many entries it holds; a section you have no entries for stays hidden.

Navigating the menu

• Click a section (or highlight it and press →) to drill into just that section's entries.
• Press ←, click the section header, or hit Esc to step back out (Esc again closes the menu).
• Type to filter — at the top level your query searches across all sections at once; inside a section it filters that section only.
• Press Enter to select.

Selecting a persona applies it to the chat — pick it again to remove it (the active one reads "Active — select to remove"). Selecting a skill or prompt inserts a chip into your message; selecting an MCP entry toggles that server for the conversation.

---

Where the library lives on disk

Your skills and prompts are plain files you own:

• Prompts — any .md file in ~/SlashspaceOS/.prompts. Optional frontmatter name: and description: control how it appears in the menu; otherwise the filename is used.
• Skills — folders containing a SKILL.md file under ~/.agents/skills. This is the cross-tool agent-skills convention, so skills you already use with tools like Claude Code appear here automatically.

Want more skills? Browse the skills.sh registry and install any skill in one click from Settings → Agent Skills.

---

Using a skill or prompt

1. Type / and select an entry from the Skills section or your prompts

2. A chip is inserted into your message — you can keep typing around it

3. On send, the file's full content is injected into the message for the model, wrapped in a Skill or Prompt tag

A single referenced file over 120,000 characters errors on send. The total expanded message is capped at 400,000 characters.

If the file behind a chip has been deleted or renamed, sending fails with Missing slash command reference… — remove the chip and re-add it from the menu.

---

Creating your own prompts

1. Create a markdown file in ~/SlashspaceOS/.prompts, for example summarize-meeting.md:

   ---
   name: Summarize meeting
   description: Turn raw notes into decisions and action items
   ---
   
   Summarize the following meeting notes. List decisions made,
   open questions, and action items with owners.

2. Type / in any chat — the new prompt appears in the menu immediately, no restart needed

---

Use cases

• Repeatable workflows: a /summarize-meeting prompt you run on pasted notes every week, instead of retyping instructions.
• Shared skills across tools: the same ~/.agents/skills folders power both this app and Claude Code, so one skill library serves your whole toolkit.
• Fast conversation setup: / → drill into MCPs and toggle a server → / → Reasoning: High → / → pick a persona, all without leaving the keyboard.
• Prompt libraries in version control: keep ~/SlashspaceOS/.prompts in a git repo and sync your best prompts across machines.