lisbeth718/agent-setup
How I run Claude Code across ~90 repos as a solo builder: rules, sub-agents, workflows, lessons
ARCHETYPE
Pragmatist
A balanced, no-drama setup: some rules, some tools, nothing extreme.
Copy this rig
# review before running: this installs third-party code $ npx degit lisbeth718/agent-setup/agents ./rig-agent-setup/agents $ npx degit lisbeth718/agent-setup/commands ./rig-agent-setup/commands $ npx degit lisbeth718/agent-setup/skills ./rig-agent-setup/skills
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 lisbeth718/agent-setup/skills/brief .claude/skills/brief $ curl -fsSL --create-dirs -o .claude/agents/fixer.md https://raw.githubusercontent.com/lisbeth718/agent-setup/main/agents/fixer.md $ curl -fsSL --create-dirs -o .claude/agents/hunter.md https://raw.githubusercontent.com/lisbeth718/agent-setup/main/agents/hunter.md $ curl -fsSL --create-dirs -o .claude/agents/researcher.md https://raw.githubusercontent.com/lisbeth718/agent-setup/main/agents/researcher.md
This rig commits no guardrails. Here is the community baseline instead — the deny/ask rules most often found across all 16.2k rigs:
{
"permissions": {
"deny": [
"Read(./.env)",
"Read(./.env.*)",
"Read(**/.env)",
"Read(~/.ssh/**)",
"Bash(git push --force:*)",
"Read(**/*.pem)",
"Read(.env)",
"Read(~/.aws/**)",
"Read(**/*.key)",
"Read(**/.env.*)",
"Bash(git push --force*)",
"Bash(rm -rf:*)",
"Bash(rm -rf *)",
"Bash(sudo *)",
"Bash(sudo:*)",
"Bash(rm -rf /)",
"Bash(git push -f:*)",
"Read(.env.*)",
"Bash(git reset --hard*)",
"Bash(rm -rf /*)"
],
"ask": [
"Bash(git push:*)",
"Bash(git push *)",
"Bash(gh pr merge *)",
"Bash(git commit:*)",
"Bash(rm *)",
"Bash(gh pr merge:*)",
"Bash(gh pr create *)",
"Bash(git reset --hard:*)",
"Bash(npm publish:*)",
"Bash(git rebase *)"
]
}
} Skills (1)
Subagents (3)
| fixer model: claude-sonnet-5-5 | Scoped, mechanical edits from an explicit file list and fix spec (apply review findings, wire components, translate dictionaries). Runs on Claude Sonnet 5.5; the caller validates with tsc and build. |
| hunter model: claude-opus-5-5 | Deep read-only review / bug-hunt / opportunity-hunt agent pinned to Opus 5.5. Use for judgment-heavy sweeps over the codebase that return a ranked findings list with file:line evidence; never for edit |
| researcher model: claude-opus-5-5 | Web research agent pinned to Claude Opus 5.5. Use for deep-research subtopics and report writing that need WebSearch/WebFetch and file writes. |
Slash commands (5)
/prepublish/review/ship/sweep/triage-alert
Similar rigs
ENEmyr/brief
Interactive docs for coding agents. Agents publish research as readable sessions; humans read, annotate, and decide, then send answers back as one prompt.
Pragmatist 2.9k tok · · Q3
mivomad33/tidwyn
AI가 매일 모아 쉬운 말로 바꿔 쓴 뉴스 브리프 아카이브 (Daily Korean news briefs rewritten by AI)
Minimalist 2.1k tok · · Q0
fjerbi/claude-code-arsenal
A tested Claude Code operating system for efficient, reliable, and focused software engineering. Includes global instructions, workflows, rules, and tools designed to minimize wasted tokens, prevent hallucinations and unnecessary loops, enf
YOLO Cowboy 4.0k tok · · Q51
jaqubowsky/fleet
Runs Claude Code and pi agents in parallel on your Mac, one Docker sandbox per issue. A host agent hands out the work, watches every sandbox and brings back tested, reviewed pull requests.
Minimalist 566 tok · · Q0