| name | sp-skill-first |
| description | Use when starting any conversation or before any response — establishes that the agent must check for relevant skills before answering, asking clarifying questions, or taking action. Replaces ad-hoc reasoning with skill-first reasoning. |
| license | MIT |
| metadata | {"author":"Adapted from JZKK720/superpowers (obra/superpowers, MIT)","source":"https://github.com/JZKK720/superpowers","source_skill":"using-superpowers","version":"1.0.0"} |
Skill-First (the bootstrap)
Before any response — including clarifying questions — check for relevant skills. Even a 1% chance a skill applies means you must invoke it to check. This is not negotiable.
The rule
- Receive the user message.
- Ask: "Could any installed skill apply to this?"
- If yes (or maybe), invoke the skill via the agent runtime's
Skill tool.
- Announce briefly: "Using [skill] to [purpose]."
- If the skill has a checklist, create a TodoWrite (or equivalent) item per checklist entry.
- Follow the skill exactly.
- Respond (with or without the skill's guidance).
What counts as "applying"
- The user message contains a trigger phrase from any skill's
description field.
- The user's intent maps to a skill's purpose (debugging, planning, TDD, design review, etc.).
- The user is about to enter a state (plan mode, design mode, debug mode) where a skill is the standard.
Red flags (these thoughts mean STOP)
| Thought | Reality |
|---|
| "This is just a simple question" | Questions are tasks. Check for skills. |
| "I need more context first" | Skill check comes BEFORE clarifying questions. |
| "Let me explore the codebase first" | Skills tell you HOW to explore. Check first. |
| "I can check git/files quickly" | Files lack conversation context. Check for skills. |
| "Let me gather information first" | Skills tell you HOW to gather information. |
| "This doesn't need a formal skill" | If a skill exists, use it. |
| "I remember this skill" | Skills evolve. Read the current version. |
| "This doesn't count as a task" | Action = task. Check for skills. |
| "The skill is overkill" | Simple things become complex. Use it. |
| "I'll just do this one thing first" | Check BEFORE doing anything. |
| "I know what that means" | Knowing the concept ≠ using the skill. Invoke it. |
| "This feels productive" | Undisciplined action wastes time. Skills prevent this. |
Skill priority
When multiple skills apply, process skills come before implementation skills. The agent invokes them in this order:
- Process skills (brainstorming, debugging, planning) — establish the how.
- Domain skills (CodeGraph, CodeReview, Memory) — apply the what.
- Implementation skills (TDD, file-edit, terminal) — execute the how.
"Build X" triggers cubecloud-brainstorm first, then cubecloud-plan, then cubecloud-tdd, then the implementation skills. Not the other way around.
Cubecloud skill ecosystem (the 17 hidden flavors + 5 user-visible)
- Hidden (auto-injected):
cubecloud-skill-first (this skill), cubecloud-tone, cubecloud-economist, cubecloud-licensor, cubecloud-tdd, cubecloud-debug, cubecloud-verify, cubecloud-brainstorm, cubecloud-plan, cubecloud-execute, cubecloud-subagents, cubecloud-parallel, cubecloud-request-review, cubecloud-receive-review, cubecloud-worktree, cubecloud-finish-branch, cubecloud-write-skill.
- User-visible (in the Skills screen):
electron-best-practices, karpathy-guidelines, design-taste-frontend, typescript-expert, ui-ux-pro-max, electron-pro, cubecloud-persona, cubecloud-onboarding, cubegraph-code-intel.
- Developer-time (in
.agents/skills/, mirrored to ~/.agents/skills/): the 20 V2.6 skills — ar-autoresearch, karpathy-guidelines, 7× po-*, 3× ecc-*, 2× gbrain-*, 6× gstack-*.
Subagent note
If you were dispatched as a subagent to execute a specific task, skip this skill. The dispatcher already invoked it for you.
Source / license
Adapted from JZKK720/superpowers · using-superpowers, MIT.
