shandybayev/claude-setup
Portable personal Claude Code setup: workflow, agents, skills, harness, templates
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 shandybayev/claude-setup/agents ./rig-claude-setup/agents $ npx degit shandybayev/claude-setup/commands ./rig-claude-setup/commands $ npx degit shandybayev/claude-setup/skills ./rig-claude-setup/skills $ npx degit shandybayev/claude-setup/hooks ./rig-claude-setup/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 shandybayev/claude-setup/skills/default-workflow .claude/skills/default-workflow $ npx degit shandybayev/claude-setup/skills/multi-session .claude/skills/multi-session $ npx degit shandybayev/claude-setup/skills/new-project .claude/skills/new-project
$ curl -fsSL --create-dirs -o .claude/agents/adversarial-reviewer.md https://raw.githubusercontent.com/shandybayev/claude-setup/main/agents/adversarial-reviewer.md $ curl -fsSL --create-dirs -o .claude/agents/builder-critical.md https://raw.githubusercontent.com/shandybayev/claude-setup/main/agents/builder-critical.md $ curl -fsSL --create-dirs -o .claude/agents/builder.md https://raw.githubusercontent.com/shandybayev/claude-setup/main/agents/builder.md $ curl -fsSL --create-dirs -o .claude/agents/plan-reviewer-critical.md https://raw.githubusercontent.com/shandybayev/claude-setup/main/agents/plan-reviewer-critical.md $ curl -fsSL --create-dirs -o .claude/agents/plan-reviewer.md https://raw.githubusercontent.com/shandybayev/claude-setup/main/agents/plan-reviewer.md $ curl -fsSL --create-dirs -o .claude/agents/planner.md https://raw.githubusercontent.com/shandybayev/claude-setup/main/agents/planner.md $ curl -fsSL --create-dirs -o .claude/agents/probe.md https://raw.githubusercontent.com/shandybayev/claude-setup/main/agents/probe.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 (3)
Subagents (7)
| adversarial-reviewer model: claude-opus-5-5 | Default-workflow ADVERSARIAL REVIEWER. One per slice, over the whole diff, after the builder holds. Runs things rather than reading them, hunts tests that test nothing, and does the re-check after the |
| builder-critical model: claude-opus-5-5 | Default-workflow BUILDER for IRREVERSIBLE work only (data moves, migrations, deploy-affecting changes), on Opus 5.5. Builds exactly one slice in its own worktree from the orchestrator's brief, with ta |
| builder model: claude-sonnet-5 | Default-workflow BUILDER. Builds exactly one slice in its own worktree from the orchestrator's brief, with targeted tests and a revert table. No git, no build, no servers. |
| plan-reviewer-critical model: claude-opus-5-5 | Default-workflow PLAN REVIEWER for IRREVERSIBLE work only (data moves, migrations, deploy-affecting changes), on Opus 5.5. Read-only gap-finder over a plan or builder brief, checked against the real c |
| plan-reviewer model: claude-sonnet-5 | Default-workflow PLAN REVIEWER. Read-only gap-finder over a plan or builder brief, checked against the real code; corrects the brief. Spawned by the orchestrator before building starts. |
| planner model: claude-opus-5-5 | Default-workflow PLANNER. Writes the plan or builder brief for a large or risky slice, after researching the real code. Spawned by the orchestrator; does not build. |
| probe model: claude-sonnet-5 | Default-workflow PROBE. Browser checks and manual-checklist walks under the owner's real conditions, returning a text verdict so screenshots stay out of the main context. |
Slash commands (8)
/batch/handoff/lessons/log/new-project/register/review/status
Similar rigs
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 ·
selmakcby/claude-agents-skills
Multi-Agent Claude Code setup — 4 uzman ajan (planner · ui-agent · builder · reviewer) + skills + Next.js demo projesi. YouTube Bölüm 1 video materyalleri.
Orchestrator 1.5k tok ·
fnayou/dotfiles
Cross-platform (macOS · Arch · Debian) dotfiles managed with GNU Stow — zsh, git, Neovim, Alacritty, btop, bat, eza, Herdr, Oh My Posh, Taskfile, Claude Code. Catppuccin Macchiato.
Pragmatist 3.5k tok ·
ulfaslak/saas_tmplt
SaaS template for Claude Code agents — SvelteKit + Postgres + Drizzle skeleton with auth, multi-tenancy, and production deploy. Bundles the agent workflow (CLAUDE.md, DNA, slash commands) that drives end-to-end feature development.
Pragmatist 7.4k tok ·