click6067-ship-it/ultrapowers
A Claude Code + Codex setup: two-model adversarial planning review (no-echo scoring), auditable council logs, and cross-session memory. Built on superpowers.
ARCHETYPE
Orchestrator
A bench of specialised subagents. The main agent mostly delegates.
Copy this rig
# review before running: this installs third-party code $ npx degit click6067-ship-it/ultrapowers/agents ./rig-ultrapowers/agents $ npx degit click6067-ship-it/ultrapowers/skills ./rig-ultrapowers/skills $ npx degit click6067-ship-it/ultrapowers/hooks ./rig-ultrapowers/hooks
MCP servers are added to Claude Code at local scope; env vars are shown as YOUR_… placeholders — we never store values. Files are fetched with degit into a separate folder so you can review before merging.
$ npx degit click6067-ship-it/ultrapowers/skills/autopilot .claude/skills/autopilot $ npx degit click6067-ship-it/ultrapowers/skills/crit .claude/skills/crit $ npx degit click6067-ship-it/ultrapowers/skills/demo .claude/skills/demo $ npx degit click6067-ship-it/ultrapowers/skills/hallmark .claude/skills/hallmark $ npx degit click6067-ship-it/ultrapowers/skills/kickoff .claude/skills/kickoff $ npx degit click6067-ship-it/ultrapowers/skills/newproject .claude/skills/newproject $ npx degit click6067-ship-it/ultrapowers/skills/orca-trio .claude/skills/orca-trio $ npx degit click6067-ship-it/ultrapowers/skills/qualityloop .claude/skills/qualityloop $ npx degit click6067-ship-it/ultrapowers/skills/race .claude/skills/race $ npx degit click6067-ship-it/ultrapowers/skills/recall .claude/skills/recall $ npx degit click6067-ship-it/ultrapowers/skills/remember .claude/skills/remember $ npx degit click6067-ship-it/ultrapowers/skills/serve .claude/skills/serve $ npx degit click6067-ship-it/ultrapowers/skills/ship .claude/skills/ship $ npx degit click6067-ship-it/ultrapowers/skills/spec-decompose .claude/skills/spec-decompose $ npx degit click6067-ship-it/ultrapowers/skills/specpack .claude/skills/specpack $ npx degit click6067-ship-it/ultrapowers/skills/techreport .claude/skills/techreport $ npx degit click6067-ship-it/ultrapowers/skills/vcheck .claude/skills/vcheck
$ curl -fsSL --create-dirs -o .claude/agents/explore.md https://raw.githubusercontent.com/click6067-ship-it/ultrapowers/master/agents/explore.md $ curl -fsSL --create-dirs -o .claude/agents/judge.md https://raw.githubusercontent.com/click6067-ship-it/ultrapowers/master/agents/judge.md $ curl -fsSL --create-dirs -o .claude/agents/redteam.md https://raw.githubusercontent.com/click6067-ship-it/ultrapowers/master/agents/redteam.md $ curl -fsSL --create-dirs -o .claude/agents/researcher.md https://raw.githubusercontent.com/click6067-ship-it/ultrapowers/master/agents/researcher.md $ curl -fsSL --create-dirs -o .claude/agents/verifier.md https://raw.githubusercontent.com/click6067-ship-it/ultrapowers/master/agents/verifier.md
This rig commits no guardrails. Here is the community baseline instead — the deny/ask rules most often found across all 7,204 rigs:
{
"permissions": {
"deny": [
"Read(./.env)",
"Read(**/.env)",
"Read(~/.ssh/**)",
"Bash(rm -rf *)",
"Read(**/*.pem)",
"Bash(rm -rf /)",
"Bash(git push --force:*)",
"Bash(sudo *)",
"Read(.env)",
"Bash(rm -rf /*)",
"Read(./.env.*)",
"Read(~/.aws/**)",
"Bash(git push --force*)",
"Bash(rm -rf:*)",
"Read(**/*.key)",
"Read(**/.env.*)",
"Bash(sudo:*)",
"Bash(git reset --hard*)",
"Bash(git reset --hard:*)",
"Read(.env.*)"
],
"ask": [
"Bash(git push:*)",
"Bash(git push *)",
"Bash(git commit:*)",
"Bash(rm *)",
"Bash(rm:*)",
"Bash(git rebase *)",
"Bash(wget *)",
"Bash(npm publish:*)",
"Bash(git commit *)",
"Bash(gh pr merge *)"
]
}
} Skills (17)
autopilotcritdemohallmarkkickoffnewprojectorca-trioqualityloopracerecallrememberserveshipspec-decomposespecpacktechreportvcheck
Subagents (5)
| Explore model: haiku | Read-only search agent for broad fan-out searches — when answering means sweeping many files, directories, or naming conventions and you only need the conclusion, not the file dumps. It reads excerpts |
| judge model: sonnet | Use when qualityloop or a scoring task needs a fresh independent judge — scores a deliverable against a rubric with per-dimension justification, lists defects by severity (blocker/major/minor), return |
| redteam model: opus | Use to red-team a plan, design, or code diff in-session. Finds failure modes, wrong scope, unstated assumptions, and the simpler version being missed. Problems only, no compliments. Distinct from /kic |
| researcher model: sonnet | Use for multi-source web research and deep web crawling. Fans out searches, fetches/crawls sources, returns a synthesized, cited summary — keeping raw crawl output OUT of the main context. If the Fire |
| verifier model: sonnet | Use to adversarially verify ONE specific claim or finding before acting on it. It tries to REFUTE the claim using independent sources and returns a verdict (holds / refuted / uncertain) with evidence. |
Similar rigs
Markuysa/vibe-coding-template
Claude Code project template: five agent roles with tool-level permission boundaries, four workflow commands, and a permission baseline that keeps secrets and build noise out of context. Built on native worktrees and subagents — no third-pa
Fort Knox 1.2k tok ·
Remus3/Resin-Compute
Genshin Impact wish and pity forecaster on an absorbing Markov chain, with resin budgeting, goal planning and an Enka.Network profile client. Headless-first Python 3.11, stdlib-only at runtime, no game data vendored.
Orchestrator 9.0k tok ·
victorkzam/claude-code-setup
My daily-driver Claude Code setup: design→build→ship agentic workflow, model-routing rules, counter-model review, and fail-closed hooks. A sanitized, clone-and-use export.
Orchestrator 845 tok ·
Chillin-spree/claude-code-system-template
A lean, skills-first Claude Code template for building web apps — vanilla by default, stack-swappable, with workflow skills, agents, and a commit/push safety gate.
Fort Knox 1.5k tok ·