diff --git a/commands/create-capability.md b/commands/create-capability.md
new file mode 100644
index 0000000..4774a5d
--- /dev/null
+++ b/commands/create-capability.md
@@ -0,0 +1,123 @@
+---
+description: Create a new capability (skill, command, agent, or a cohesive set) for the architecture repository.
+argument-hint: <description>
+---
+
+# Create Capability
+
+@~/.claude/skills/capability-writing/SKILL.md
+
+Create new capabilities following established patterns. A capability may be a single component or a cohesive set (skill + command + agent).
+
+## Process
+
+1. **Understand the capability**: Analyze "$1" to understand what the user wants to build
+   - What domain or workflow does this cover?
+   - What user need does it address?
+   - What existing capabilities might overlap?
+
+2. **Determine components needed**: Based on the description, recommend which components:
+
+   | Pattern | When to Use |
+   |---------|-------------|
+   | Skill only | Knowledge to apply automatically (reused across commands) |
+   | Command only | User-invoked workflow using existing skills |
+   | Command + Skill | New knowledge + workflow to use it |
+   | Command + Agent | Workflow with isolated worker for complex subtasks |
+   | Full set | New domain expertise + workflow + isolated work |
+
+   Present recommendation with reasoning:
+   ```
+   ## Recommended Components for: $1
+
+   Based on your description, I recommend:
+   - **Skill**: `name` - [why this knowledge is needed]
+   - **Command**: `/name` - [what user action this enables]
+   - **Agent**: `name` - [why isolation/specialization is needed] (optional)
+
+   Reasoning: [explain why this combination fits the need]
+   ```
+
+3. **Gather information**: For each recommended component, ask:
+
+   **For all components:**
+   - Name (kebab-case, descriptive)
+   - Description (one-line summary)
+
+   **For Skills:**
+   - What domain/knowledge does this cover?
+   - What are the key concepts to teach?
+   - What patterns or templates should it include?
+
+   **For Commands:**
+   - What triggers this workflow?
+   - What are the key steps?
+   - What approval points are needed?
+   - What skills should it reference?
+
+   **For Agents:**
+   - What specialized role does this fill?
+   - What skills does it need?
+   - Should it be read-only (no Edit/Write)?
+
+4. **Select appropriate models**:
+
+   | Model | Use For |
+   |-------|---------|
+   | `haiku` | Simple fetch/display commands, formatting tasks |
+   | `sonnet` | Most commands and agents (default) |
+   | `opus` | Deep reasoning, architectural analysis, complex judgment |
+
+   For each component, recommend a model with reasoning.
+
+5. **Generate files**: Create content using templates from capability-writing skill
+
+   Ensure proper inter-references:
+   - Command references skill via `@~/.claude/skills/name/SKILL.md`
+   - Agent lists skills in `skills:` frontmatter (names only, not paths)
+   - Command spawns agent via Task tool if agent is part of the set
+
+6. **Present for approval**: Show all generated files with their full content:
+   ```
+   ## Generated Files
+
+   ### skills/name/SKILL.md
+   [full content]
+
+   ### commands/name.md
+   [full content]
+
+   ### agents/name/AGENT.md (if applicable)
+   [full content]
+
+   Ready to create these files?
+   ```
+
+7. **Create files** in correct locations after approval:
+   - `skills/<name>/SKILL.md`
+   - `commands/<name>.md`
+   - `agents/<name>/AGENT.md`
+
+8. **Report success**:
+   ```
+   ## Capability Created: name
+
+   Files created:
+   - skills/name/SKILL.md
+   - commands/name.md
+   - agents/name/AGENT.md (if applicable)
+
+   Next steps:
+   1. Run `make install` to symlink to ~/.claude/
+   2. Test with: /name (for commands)
+   3. Skills will auto-activate based on context
+   ```
+
+## Guidelines
+
+- Follow all conventions from capability-writing skill
+- Reference existing skills rather than duplicating knowledge
+- Keep components focused - split if scope is too broad
+- Commands should have approval checkpoints before significant actions
+- Default to `sonnet` model unless there's a clear reason for haiku/opus
+- Skills should have descriptive `description` fields for auto-activation