This is the Bonsai Software fork of obra/superpowers, customized with team-specific skills and Tier 2 orchestration edits. Team members should install from this repo, not upstream.
In Claude Code:
/plugin marketplace add BonsaiSoftware/bonsaipowers
/plugin install bonsaipowers@bonsaipowers
The first command registers this repo as a marketplace; the second installs the plugin from it. The default branch on GitHub is bonsai-custom — that's where all Bonsai customization lives — so the marketplace fetches our fork, not upstream.
To update:
/plugin update bonsaipowers
After installing, start a new session and say brainstorm a small feature. The first checklist step should be "Structured codebase recon" (step 1) — the agent should grep package.json, read CLAUDE.md, and find a similar existing feature before asking any clarifying questions. Later, before proposing approaches, it should run at least one context7 query AND one WebSearch query (the Tier 2 research HARD-GATE). If those Bonsai-specific behaviors are missing, you're on the wrong branch.
The Bonsai workflow depends on three MCP servers that are not bundled with the plugin — per the "no MCP in plugin" rule (see CLAUDE.md), MCPs belong in the consuming project's .mcp.json (shared with the team, committed to git) or the user's ~/.claude.json (personal, not shared). The plugin ships zero MCP config so it works cleanly with both per-project and per-user setups.
| MCP | Used by | Why it's required |
|---|---|---|
context7 |
brainstorming (step 4 research HARD-GATE), writing-plans (research section), executing-plans (JIT API verification), subagent-driven-development (implementer prompts) |
Library documentation and API signature lookups. Without it, brainstorming fails its research gate, plans ship guessed method signatures, and implementers can't verify APIs before writing code. |
microsoft-docs |
writing-plans (research step 6, Azure features) |
Official Azure and Microsoft documentation plus code samples via microsoft_code_sample_search. Required whenever a plan touches an Azure service — which, for this team, is most of them. |
chrome-devtools |
bonsai-verify-ui |
Drives real Chrome from the orchestrator's main context (mcp__chrome-devtools__navigate_page, click, fill, take_snapshot, evaluate_script, etc.). Without it, runtime UI verification after subagent-driven-development cannot run. |
Follow each MCP server's own install instructions. If any are missing, the relevant skills will surface an explicit error rather than silently degrading — but it's a lot faster to configure all three up front.
For critical work where you don't want surprise updates, pin to a tag by passing it as a ref when adding the marketplace:
/plugin marketplace add BonsaiSoftware/[email protected]
/plugin install bonsaipowers@bonsaipowers
Current version lives in .claude-plugin/plugin.json. Maintainers tag stable points on bonsai-custom when significant Tier 2 changes ship.
If you're editing this repo (not just installing it), read CLAUDE.md and docs/bonsai/customizing/README.md before touching anything in skills/ — the three-tier model determines what's safe to change. Local development loop is in the Local Development section further down.
Superpowers is a complete software development methodology for your coding agents, built on top of a set of composable skills and some initial instructions that make sure your agent uses them.
Give your agent Superpowers: Claude Code, Codex CLI, Codex App, Factory Droid, Gemini CLI, OpenCode, Cursor, GitHub Copilot CLI.
It starts from the moment you fire up your coding agent. As soon as it sees that you're building something, it doesn't just jump into trying to write code. Instead, it steps back and asks you what you're really trying to do.
Once it's teased a spec out of the conversation, it shows it to you in chunks short enough to actually read and digest.
After you've signed off on the design, your agent puts together an implementation plan that's clear enough for an enthusiastic junior engineer with poor taste, no judgement, no project context, and an aversion to testing to follow. It emphasizes true red/green TDD, YAGNI (You Aren't Gonna Need It), and DRY.
Next up, once you say "go", it launches a subagent-driven-development process, having agents work through each engineering task, inspecting and reviewing their work, and continuing forward. It's not uncommon for Claude to be able to work autonomously for a couple hours at a time without deviating from the plan you put together.
There's a bunch more to it, but that's the core of the system. And because the skills trigger automatically, you don't need to do anything special. Your coding agent just has Superpowers.
If Superpowers has helped you do stuff that makes money and you are so inclined, I'd greatly appreciate it if you'd consider sponsoring my opensource work.
Thanks!
- Jesse
Installation differs by harness. If you use more than one, install Superpowers separately for each one.
Superpowers is available via the official Claude plugin marketplace
-
Install the plugin from Anthropic's official marketplace:
/plugin install superpowers@claude-plugins-official
The Superpowers marketplace provides Superpowers and some other related plugins for Claude Code.
-
Register the marketplace:
/plugin marketplace add obra/superpowers-marketplace
-
Install the plugin from this marketplace:
/plugin install superpowers@superpowers-marketplace
Superpowers is available via the official Codex plugin marketplace.
-
Open the plugin search interface:
/plugins
-
Search for Superpowers:
superpowers
-
Select
Install Plugin.
Superpowers is available via the official Codex plugin marketplace.
- In the Codex app, click on Plugins in the sidebar.
- You should see
Superpowersin the Coding section. - Click the
+next to Superpowers and follow the prompts.
-
Register the marketplace:
droid plugin marketplace add https://github.com/obra/superpowers
-
Install the plugin:
droid plugin install superpowers@superpowers
-
Install the extension:
gemini extensions install https://github.com/obra/superpowers
-
Update later:
gemini extensions update superpowers
OpenCode uses its own plugin install; install Superpowers separately even if you already use it in another harness.
-
Tell OpenCode:
Fetch and follow instructions from https://raw.githubusercontent.com/obra/superpowers/refs/heads/main/.opencode/INSTALL.md -
Detailed docs: docs/README.opencode.md
-
In Cursor Agent chat, install from marketplace:
/add-plugin superpowers -
Or search for "superpowers" in the plugin marketplace.
-
Register the marketplace:
copilot plugin marketplace add obra/superpowers-marketplace
-
Install the plugin:
copilot plugin install superpowers@superpowers-marketplace
If you've cloned this fork and want to test changes without pushing to a remote or installing through a marketplace, use Claude Code's --plugin-dir flag. This loads the plugin directly from your local clone for the current session only — no persistent install, no marketplace setup.
claude --plugin-dir /Volumes/corsair-ex/bonsai-git/bonsaipowersSession-scoped: this is NOT a persistent install. Every time you launch claude without the flag, the plugin is not loaded. To make it ergonomic, add a shell alias:
alias claudebonsai='claude --plugin-dir /Volumes/corsair-ex/bonsai-git/bonsaipowers'Iteration loop:
- Edit any file in the repo (
skills/,agents/,commands/,hooks/, docs, etc.) - Run
/reload-pluginsinside Claude Code to pick up the change — no restart needed - Test the change in the same session
Smoke test that the plugin is loaded correctly:
- Run
/help— you should see skills listed under thebonsaipowersnamespace (e.g.,/bonsaipowers:brainstorming,/bonsaipowers:writing-plans) - Say
brainstorm a small featurein the chat — the agent's first checklist step should be "Structured codebase recon" (step 1, a Bonsai Tier 2 addition), and before proposing approaches it should run bothcontext7andWebSearchqueries (the step-4 HARD-GATE). If upstream behavior leaks through (generic "Explore project context" instead of the targeted recon, or no research queries), the Tier 2 edits aren't loaded. Seedocs/bonsai/tier-2-edits.md.
Collision with upstream superpowers: if you also have obra's superpowers plugin installed (either globally or via a marketplace), both will load and you'll see both sets of skills in /help. Because this fork is renamed to bonsaipowers, they coexist without namespace conflicts, but the duplicate skill list can be confusing. Uninstall the upstream copy while developing here if it gets in the way:
/plugin uninstall superpowers
For full maintainer documentation (three-tier skill model, upstream sync workflow, Tier 2 edit tracking), see docs/bonsai/customizing/README.md and CLAUDE.md.
-
brainstorming - Activates before writing code. Refines rough ideas through questions, explores alternatives, presents design in sections for validation. Saves design document.
-
using-git-worktrees - Activates after design approval. Creates isolated workspace on new branch, runs project setup, verifies clean test baseline.
-
writing-plans - Activates with approved design. Breaks work into bite-sized tasks (2-5 minutes each). Every task has exact file paths, complete code, verification steps.
-
subagent-driven-development or executing-plans - Activates with plan. Dispatches fresh subagent per task with two-stage review (spec compliance, then code quality), or executes in batches with human checkpoints.
-
test-driven-development - Activates during implementation. Enforces RED-GREEN-REFACTOR: write failing test, watch it fail, write minimal code, watch it pass, commit. Deletes code written before tests.
-
requesting-code-review - Activates between tasks. Reviews against plan, reports issues by severity. Critical issues block progress.
-
finishing-a-development-branch - Activates when tasks complete. Verifies tests, presents options (merge/PR/keep/discard), cleans up worktree.
The agent checks for relevant skills before any task. Mandatory workflows, not suggestions.
Testing
- test-driven-development - RED-GREEN-REFACTOR cycle (includes testing anti-patterns reference)
Debugging
- systematic-debugging - 4-phase root cause process (includes root-cause-tracing, defense-in-depth, condition-based-waiting techniques)
- verification-before-completion - Ensure it's actually fixed
Collaboration
- brainstorming - Socratic design refinement
- writing-plans - Detailed implementation plans
- executing-plans - Batch execution with checkpoints
- dispatching-parallel-agents - Concurrent subagent workflows
- requesting-code-review - Pre-review checklist
- receiving-code-review - Responding to feedback
- using-git-worktrees - Parallel development branches
- finishing-a-development-branch - Merge/PR decision workflow
- subagent-driven-development - Fast iteration with two-stage review (spec compliance, then code quality)
Meta
- writing-skills - Create new skills following best practices (includes testing methodology)
- using-superpowers - Introduction to the skills system
- Test-Driven Development - Write tests first, always
- Systematic over ad-hoc - Process over guessing
- Complexity reduction - Simplicity as primary goal
- Evidence over claims - Verify before declaring success
Read the original release announcement.
The general contribution process for Superpowers is below. Keep in mind that we don't generally accept contributions of new skills and that any updates to skills must work across all of the coding agents we support.
- Fork the repository
- Switch to the 'dev' branch
- Create a branch for your work
- Follow the
writing-skillsskill for creating and testing new and modified skills - Submit a PR, being sure to fill in the pull request template.
See skills/writing-skills/SKILL.md for the complete guide.
Superpowers updates are somewhat coding-agent dependent, but are often automatic.
MIT License - see LICENSE file for details
Superpowers is built by Jesse Vincent and the rest of the folks at Prime Radiant.
- Discord: Join us for community support, questions, and sharing what you're building with Superpowers
- Issues: https://github.com/obra/superpowers/issues
- Release announcements: Sign up to get notified about new versions