super-board

Drag a card into Ready, walk away, come back to a merged PR with evidence: 7 skills — 5 primary, 2 secondary — 9 commands, 8 guards.

Install

curl -fsSL https://raw.githubusercontent.com/EricTechPro/super-board/main/get.sh | bash   # in your project folder

🧩 super-board installer
🔍 checking what you need
   ✓ curl · tar · python3 · gh · jq · node
📦 downloading EricTechPro/[email protected]
🔧 installing
   ✓ skills, scripts and workflows → .claude/
   🛡️  6 guard hooks → .claude/settings.json (backup kept)
🧠 helper skills: installed
🎉 super-board 3.0.0 is installed
👉 next: open Claude Code here and run /super-board onboard

Or from a checkout: ./install.sh [--no-hooks] [--protect-main] /path/to/your-project. Then, inside
Claude Code: /super-board onboard (8 steps, Enter takes the recommended answer), move cards to
Ready, /super-board run <slug>.

Skills

Primary — the board and its lanes

Skill	What it does
`/super-board`	Orchestrator: plans waves, runs Build → QA → Review, merges when green.
`/super-build`	Builder lane: worktree, smallest safe change, tests, draft PR.
`/super-qa`	Tester lane: evidence, test-gap check, screenshots, or bounce to Build.
`/super-review`	Reviewer lane: own hypotheses, remembers prior findings, merge gate.
`/super-collect`	Finds problems (Sentry, PostHog, issues, PRs, architecture) and files verified Backlog cards.

Secondary — standalone helpers

Skill	What it does
`/visual`	One HTML page: branch recap, plan, or codebase map with diagrams.
`/ui-refine-loop`	Critique → refine loop that polishes one page or component.

Each name links to that skill's README. Lane skills run as agents inside the super-board-wave workflow.

Commands


`/super-board onboard`	8-step setup: 🔍 Checks · 🔑 GitHub · 🗂️ Board · 🌿 Branch · 📜 AGENTS.md · 🛡️ Policies · 📥 Bug sources · ✅ Review; re-run resumes or upgrades
`/super-board lint`	Flags vague ACs and unreadable `Blocked by` lines before agents spend tokens
`/super-board status`	Read-only board snapshot, column counts, in-flight work
`/super-board run <slug> [--low\|--high]`	The loop, until the board drains or a halt gate fires; also resumes
`/super-board stop`	Posts "stopped mid-flight" notes, releases claims, stops workers
`/super-collect [sentry\|posthog\|github\|prs\|architecture\|<custom>] [--since 30d]`	Finds problems, verifies each, files them into Backlog (dry-run first)
`/ui-refine-loop <route>`	Polishes one page or component in critique → refine rounds
`/visual [recap\|plan\|<path>]`	One HTML page of a branch, a plan, or part of the codebase

Guards

Run automatically once installed. Python stdlib, JSON in, JSON out.


guard-worktree-path	Blocks `git worktree add` outside `.claude/worktrees/`
guard-secrets	Blocks reading or piping dotenv files, SSH keys and credential files; key names via `super-board-env-check.sh`
guard-key-literals	Blocks a live-looking API key written into a file; flags one already there
guard-delete-outside	Blocks `rm`, `find -delete` and `git clean` aimed outside the project, `~` or `/`
guard-protected-push	Opt-in (`onboard` asks, or `install.sh --protect-main`): blocks direct and force pushes to main/master/base
README sync	Regenerates this skill table; pre-commit and PostToolUse hooks keep it fresh
cleanup-wt	Removes merged worktrees and branches after each merge and at session start, with a recovery file
merge gate	Merges only after the current base plus your `verify_commands` pass, pinned to the reviewed commit; applies `merge_policy` and runs allowed DB migrations

Setup notes

Needs Claude Code, gh (authenticated for the board's owner), jq, bash 3.2+ and Python 3.
The board is a GitHub Project (v2) with a Status field: Backlog, Ready, Building, QA, Review, Blocked, Done. onboard reuses your best-matching board (adding what is missing, keeping every card) or creates one.
Three labels route cards: qa skips Building (test what exists), bug and feature are built first; no label is built too.
Default backend is in-session dynamic workflows: turn them on in /config. Headless claude -p is opt-in (worker_backend: "claude-p").
Set verify_commands in the config. Without them the merge gate cannot prove the result builds, and says so.
onboard fixes what the board needs without asking (skills, scripts, workflows, guards, settings, old folders, config keys) and asks only to install a system tool or sign in. An older super-board is upgraded in place (see Upgrading from 2.x).
onboard asks once what the robot may do: auto-merge normal changes up to 400 changed lines (money, auth, destructive schema like DROP/TRUNCATE/RENAME, and bigger PRs wait for you; additive migrations run only on the databases you allow, never live), protect main, and which databases it may migrate (default test + staging). Anything only you may run lands in Blocked tagged 🙋 with the exact command; comment done and the next wave merges.
Auto-merge needs the merge-gate and gh pr merge lines in your allowlist; onboard offers them as a diff.
onboard can make AGENTS.md the source of truth (CLAUDE.md becomes @AGENTS.md) and keeps a managed super-board section in it. Plugin install: the command is /super-board:super-board onboard, and its 🔍 Checks step installs the missing scripts, workflows and hooks.
Bug sources take any extra source in onboard — a link, app name, API URL, MCP server or command — proven readable (read-only) before it is saved; /super-collect runs it like the built-ins.
Testing a live site with no repo is not a board: /super-qa <url> runs on its own.
To enable the usage check, add printf '%s' "$input" | bash <repo>/.claude/bin/super-board-usage.sh record to your status-line script.
Cards need acceptance criteria: QA grades against them, and lint tells you which are missing.

Learn more

Credits

Designed and maintained by Eric Tech. MIT, see LICENSE.
Skill structure inspired by obra/superpowers.
Lanes run on the mattpocock/skills process stack.
super-collect and visual adapt BuilderIO/skills (MIT); visual's diagrams follow tt-a1i/archify (MIT).
ui-refine-loop, the cleanup-wt hook and guard-worktree-path come from Eric Tech's BookKeepingApp.

Skills load on a description match. Commands you type. Guards run on their tool events. Adding a skill? Write its SKILL.md and README.md, add a line to skills/families.json, and the README updates itself.