Skip to content

Quickstart

Five minutes from "never installed it" to "shipped a real change with it."

1. Install

In Claude Code:

/plugin marketplace add ZunoSmartLabs/zsl-superpowers
/plugin install zsl@zsl-superpowers

Restart Claude Code. The skills now surface as /zsl:<skill-name>.

2. Try the smallest skill first

Open Claude Code in any repo and run:

/zsl:grill-me

Tell it about something you're stuck on. It'll interview you until your decision tree is fully resolved — no code changes, no commits, just a conversation that forces clarity. This is the cheapest way to feel the difference: the agent stops guessing and starts asking.

You can also try /zsl:caveman for a token-cut compressed reply mode, or /zsl:timesheet to turn this morning's sessions into standup notes.

3. Configure a repo

For the engineering skills (tdd, triage, to-prd, to-issues, improve-codebase-architecture, diagnose, zoom-out) you need to teach the plugin about this repo's setup. From inside the repo, run:

/zsl:setup-zsl-superpowers

It asks four questions:

  1. Issue tracker — GitHub, GitLab, or local markdown files under .scratch/.
  2. Triage label vocabulary — the actual label strings you use (the plugin maps these to canonical roles like ready-for-agent).
  3. Where domain docs live — CONTEXT.md, ADRs path, etc.
  4. Ship style — PR-style or direct-push.

The output lands in AGENTS.md (or CLAUDE.md), docs/agents/, and (for PR-style repos) the project board mapping at docs/agents/project-board.md.

Full setup details →

4. Ship one thing through the loop

Pick something small — a one-day feature, a bug you've been putting off. Then:

  1. /zsl:grill-with-docs — interview, sharpens terminology, updates CONTEXT.md and ADRs as decisions crystallise.
  2. /zsl:to-prd — write the conversation up as a PRD on your issue tracker.
  3. /zsl:to-issues — slice the PRD into vertical-slice sub-issues with explicit Blocked by graphs.
  4. /zsl:triage — move each child to ready-for-agent with an agent brief.
  5. /zsl:tdd-parallel — fan the AFK slices out into worktrees in waves and open one integration PR. (PR-style repos only — see the deep-dive.)

You'll feel the shift around step 3: instead of an agent diving in flailing, you're handed a slicing plan you can edit before any code gets written.

The full loop →

  • The end-to-end loop — how the skills compose, with state-machine diagrams.
  • Parallel TDD deep-dive — why one PR, what an integration PR looks like.
  • Skills — every skill, with its trigger phrases.
  • FAQ — Codex/Cursor compatibility, telemetry, opt-in subset, updates.