pahlevikun/levistack
Agent skills for people tired of AI slop. 35 skills, 11 subagents, 41 rules and 5 hooks for Claude Code, Cursor and Codex. Verifies before it says done.
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 pahlevikun/levistack/.claude ./rig-levistack # inspect, then merge into .claude/ 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 pahlevikun/levistack/.cursor/skills/manage-stack .claude/skills/manage-stack $ npx degit pahlevikun/levistack/skills/engineering/super-noslop/skills/super-noslop-code .claude/skills/super-noslop-code
$ curl -fsSL --create-dirs -o .claude/agents/bulk-reader.md https://raw.githubusercontent.com/pahlevikun/levistack/main/agents/bulk-reader.md $ curl -fsSL --create-dirs -o .claude/agents/context-harvester.md https://raw.githubusercontent.com/pahlevikun/levistack/main/agents/context-harvester.md $ curl -fsSL --create-dirs -o .claude/agents/coordinator.md https://raw.githubusercontent.<redacted>.md $ curl -fsSL --create-dirs -o .claude/agents/explorer.md https://raw.githubusercontent.com/pahlevikun/levistack/main/agents/explorer.md $ curl -fsSL --create-dirs -o .claude/agents/implementer.md https://raw.githubusercontent.<redacted>.md $ curl -fsSL --create-dirs -o .claude/agents/planner.md https://raw.githubusercontent.com/pahlevikun/levistack/main/agents/planner.md $ curl -fsSL --create-dirs -o .claude/agents/reviewer.md https://raw.githubusercontent.com/pahlevikun/levistack/main/agents/reviewer.md $ curl -fsSL --create-dirs -o .claude/agents/scaffolder.md https://raw.githubusercontent.com/pahlevikun/levistack/main/agents/scaffolder.md $ curl -fsSL --create-dirs -o .claude/agents/security-reviewer.md https://raw.githubusercontent.com/pahlevikun/levistack/main/agents/security-reviewer.md $ curl -fsSL --create-dirs -o .claude/agents/tdd-runner.md https://raw.githubusercontent.com/pahlevikun/levistack/main/agents/tdd-runner.md $ curl -fsSL --create-dirs -o .claude/agents/template-writer.md https://raw.githubusercontent.com/pahlevikun/levistack/main/agents/template-writer.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 (2)
Subagents (11)
| bulk-reader model: haiku | Use when a large file read would flood the caller's context, or to learn a sibling pattern cheaply. Reads files for a question and returns structured bullets only, never contents. |
| context-harvester | Use before non-trivial work to search project knowledge (AGENTS.md, rules, docs, ADRs, notes, changelog) and return a compact summary with sources, instead of dumping docs into the main conversation. |
| coordinator | Use to turn an agreed plan into task cards, route each slice to the right agent, evaluate results and escalate. Also has a route-only mode that returns a JSON plan of which agents and context a task n |
| explorer | Use to find where code lives or to answer one bounded question about the codebase, docs or git history. Read-only; returns path:line findings, never file dumps. |
| implementer | Use for bounded code edits on paths the caller owns: add a field, fix a handler, wire a client, extend tests, change tooling. For net-new structure use scaffolder first; run reviewer after non-trivial |
| planner | Use when a task needs a dedicated planning pass: weigh approaches, write or revise the plan with evals and open decisions. Plans only; never implements. |
| reviewer | Use after non-trivial edits and before merge: scores the work against its evals, then reviews the diff for silent failures, principle violations, missing tests and scope creep. |
| scaffolder | Use to add a new module, package, endpoint or feature skeleton from the project's templates and wire it into registration. Produces a compiling skeleton; business logic is left for implementer. |
| security-reviewer | Use on diffs touching auth, PII, secrets, webhooks, spawned processes, external clients or concurrency. STRIDE-minded: traces data flow end to end and cites file:line evidence for every finding. |
| tdd-runner | Use when implementing or fixing a bounded slice test-first: runs the scoped tests, reads the real failure, makes the minimal fix, never weakens assertions. |
| template-writer model: haiku | Use for purely mechanical file generation: given template files, a placeholder map and destination paths, write the substituted files. Never decides shape or logic. |
Slash commands (1)
/handoff
Cursor rules (5)
commentscommit-messages · alwaysgit-workflow · alwaysmanage-stack · alwaysno-worktrees · always
Similar rigs
ingenierodeatascos-cyber/ai-project-template
AI-First project template for building software with orchestrated AI agents using Spec-Driven Development. Includes router, planner, implementer and reviewer workflows for structured development with Codex, Claude Code or similar AI coding
Orchestrator 196 tok ·
BillDX/spicy-to-dark-factory-starter
Workshop starter for 'From Spicy Autocomplete to Dark Factory: Agentic Coding in Practice' — a FastAPI mini-app with a full Claude Code harness (agents, slash commands, hooks, sandbox settings) and self-paced exercises.
Orchestrator 706 tok ·
Matheuscara/claude-code
🤖 Config do meu Claude Code para homelab: regras, hooks de segurança, subagentes, skills, commands e instincts (cherry-pick inspirado no ECC)
Pragmatist 1.6k tok ·
orynvn/claude-workspace-setup
A project template for Claude Code CLI — opinionated, token-efficient, and built for multi-agent workflows.
Orchestrator 22.3k tok ·