AI agent orchestration, quality hooks, and smart routing for Claude Code
npx claudepluginhub speson/not-my-reforgenot oh-my. reforged. — AI agent orchestration, quality hooks, and smart routing for Claude Code
speson
not oh-my. reforged.
한국어 | English
Your Claude Code shouldn't need a babysitter.
not-my-reforge teaches your AI agent to earn trust, fix its own mistakes, and orchestrate itself — so you can stop micromanaging prompts and start shipping.
oh-my-* gave agents commands. We gave them autonomy.
Most Claude Code plugins give you more buttons to press. We took the opposite approach: fewer decisions for you, better decisions by the agent.
Zero runtime dependencies. Pure hooks architecture. No setup wizard. No config ceremony.
Just install and let it cook.
/plugin add marketplace https://github.com/speson/not-my-reforge
/plugin install not-my-reforge
For the best experience (sidebar dashboard, parallel execution), run inside tmux:
tmux new-session "claude --dangerously-skip-permissions"
git clone https://github.com/speson/not-my-reforge
cd not-my-reforge && npm install && npm run build
# Run inside tmux for full feature support
tmux new-session "claude --plugin-dir ."
Note: When developing this plugin itself, always use
--plugin-dirto point at the local source. The installed version (~/.claude/plugins/cache/...) is a separate copy — edits to the repo won't take effect until you reinstall or use this flag. Using--plugin-diralso ensures hooks run against your latest build, avoiding stale-cache issues like the verification gate running outdated code.
#orch implement auth, profiles, and notifications # Smart orchestration
#deep refactor the authentication module # Deep analysis (opus)
#quick what does this function do? # Fast answer (haiku)
#sec audit the API authentication layer # Security scan
#score # Quality score report
#trust # Trust level status
/not-my-reforge:diff-review # 5-perspective review
#search UserService # DeepSearch
8 modes — or let #orch auto-select:
| Mode | Command | Best For |
|---|---|---|
| Ralph Loop | reforge loop <task> | Iterative refinement until success |
| Autopilot | reforge autopilot <goal> | Sequential multi-task automation |
| Pipeline | reforge pipeline <goal> | Quality gates: plan → implement → verify → fix → review |
| Team | reforge team N <task> | Parallel workers + worktree isolation |
| Swarm | reforge swarm <goal> | Parallel research and analysis |
| QA Loop | reforge qa <target> | Auto test → fix → retest cycle |
| Ralplan | reforge ralplan <goal> | Iterative consensus planning |
| Ultrawork | #ultrawork <task> / #uw <task> | ALL agents in maximum parallel execution |
#orch routing:
| Pattern | Mode |
|---|---|
| Multiple tasks (comma/and) | team |
| Multi-task + complex scope | ultrawork |
| "fix all" + test-related | qa |
"fix all" / --verify | ralph |
| Analysis / research / audit | swarm |
| Design / plan / architect | ralplan |
| Quality-critical (payment, auth) | pipeline |
reforge cancel # cancel all modes
/not-my-reforge:mode # check active mode
20 Core + 5 Team agents:
Claude Code marketplace entries for the plugin-safe Antigravity Awesome Skills library and its compatible editorial bundles.
Lazy senior dev mode for AI agents. The best code is the code you never wrote.
Production-ready workflow orchestration with 84 marketplace plugins, 192 local specialized agents, and 156 local skills - optimized for granular installation and minimal token usage