alsharifabdull/shipmiracle
Local, safety-first workspace for AI coding agents: plan, build, test and review, with every step that matters waiting for your approval. macOS first.
ARCHETYPE
Orchestrator
A bench of specialised subagents. The main agent mostly delegates.
GUARDRAILS
3/5
Blocks destructive commands · Pre-tool screening hook · No YOLO mode · details
Quality 33/100 Guardrails v2
Copy this rig
# review before running: this installs third-party code
$ npx degit alsharifabdull/shipmiracle/templates/base/.claude ./rig-shipmiracle # 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.
$ curl -fsSL --create-dirs -o .claude/agents/builder-backend.md https://raw.githubusercontent.<redacted>-backend.md $ curl -fsSL --create-dirs -o .claude/agents/builder-frontend.md https://raw.githubusercontent.<redacted>-frontend.md $ curl -fsSL --create-dirs -o .claude/agents/builder-ops.md https://raw.githubusercontent.<redacted>-ops.md $ curl -fsSL --create-dirs -o .claude/agents/builder.md https://raw.githubusercontent.<redacted>.md $ curl -fsSL --create-dirs -o .claude/agents/designer.md https://raw.githubusercontent.<redacted>.md $ curl -fsSL --create-dirs -o .claude/agents/planner.md https://raw.githubusercontent.<redacted>.md $ curl -fsSL --create-dirs -o .claude/agents/reporter.md https://raw.githubusercontent.<redacted>.md $ curl -fsSL --create-dirs -o .claude/agents/reviewer.md https://raw.githubusercontent.<redacted>.md $ curl -fsSL --create-dirs -o .claude/agents/scoper.md https://raw.githubusercontent.<redacted>.md $ curl -fsSL --create-dirs -o .claude/agents/verifier.md https://raw.githubusercontent.<redacted>.md
Merge into .claude/settings.json (project) or ~/.claude/settings.json (user). Hook commands reference scripts in the source repo — copy those too.
{
"hooks": {
"PreToolUse": [
{
"matcher": "Bash",
"hooks": [
{
"type": "command",
"command": "$CLAUDE_PROJECT_DIR/.claude/hooks/block-secrets.sh"
}
]
},
{
"matcher": "*",
"hooks": [
{
"type": "command",
"command": "$CLAUDE_PROJECT_DIR/.claude/hooks/shipm-state.sh working"
}
]
}
]
}
} Subagents (10)
| builder-backend model: sonnet | Use when services, APIs, data, integrations, imports or migrations change; validates input, fails closed, reversible migrations, idempotent imports. Not for UI or deploys. |
| builder-frontend model: sonnet | Use when UI code changes (web/iOS screens, components, styles, a11y, RTL); screenshot before done. Not for APIs, data, deploys. |
| builder-ops model: sonnet | Use when CI, build, launchd, Docker, infra or deploy config changes; dry-run first; production deploy is the owner's call. Not for app features. |
| builder model: sonnet | Use when the change is docs, content, research write-ups, notes, scripts or config text, following an approved plan. Not for UI (builder-frontend), services/data (builder-backend), CI/deploy (builder- |
| designer model: sonnet | Use when UI is about to be built or changed, to write the spec, states and DESIGN.md text, or to review built UI against it. Not for app code. |
| planner model: opus | Use when any change needs a design/plan or a mid-work impact check; returns one recommendation + tasks with acceptance criteria. Not for writing code or files. |
| reporter model: haiku | Use when the report queue hands you a facts JSON file, to turn it into at most 3 summary lines without adding facts. Not for anything else. |
| reviewer model: opus | Use when a material change is done and before it is reported done; the caller names the lens, correctness (logic, tests, regressions) or risk (security, privacy, data loss, auth). One verdict per lens |
| scoper model: sonnet | Use when a request needs acceptance criteria, edge cases, open questions before planning. Not for plans, code or tests. |
| verifier model: sonnet | Use when a build is finished and before it is called done, to run the real tests, map criteria to evidence and give PASS or FAIL. May add or fix tests under test paths only. Not for product code. |
Hooks (10)
| event | matcher | runs |
|---|---|---|
| SessionStart | * | $CLAUDE_PROJECT_DIR/.claude/hooks/session-start.sh |
| PreToolUse | Bash | $CLAUDE_PROJECT_DIR/.claude/hooks/block-secrets.sh |
| PreToolUse | * | $CLAUDE_PROJECT_DIR/.claude/hooks/shipm-state.sh working |
| PostToolUse | Edit|Write | $CLAUDE_PROJECT_DIR/.claude/hooks/format.sh |
| PostToolUse | ExitPlanMode | $CLAUDE_PROJECT_DIR/.claude/hooks/save-plan.sh |
| PreModelSwitch | * | $CLAUDE_PROJECT_DIR/.claude/hooks/model-switch.sh |
| Stop | * | $CLAUDE_PROJECT_DIR/.claude/hooks/stop-gate.sh |
| Stop | * | $CLAUDE_PROJECT_DIR/.claude/hooks/shipm-state.sh done |
| UserPromptSubmit | * | $CLAUDE_PROJECT_DIR/.claude/hooks/shipm-state.sh working |
| Notification | * | $CLAUDE_PROJECT_DIR/.claude/hooks/shipm-state.sh needs-input |
Slash commands (1)
/feature
Similar rigs
iuliandita/agents
Global prompts and subagent roles for coding agents, deployed to every harness you use.
Orchestrator 1.5k tok · · Q3
joshuakatt/Bob-The-Builder
Autonomous task orchestrator for Kiro + Claude specs.
YOLO Cowboy 1.4k tok · · Q0
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 11.3k tok · · Q51
Remus3/ClockSpeed
Local, deterministic production planner and advisor for Satisfactory. Reads your own game install and save file instead of a wiki dump, so answers match the patch you are actually playing. Exact rational arithmetic, no language model in the
Orchestrator 10.6k tok · · Q51