Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/anshss/shiploop/setupgit clone --depth 1 https://github.com/anshss/shiploopWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00024 | $0.07038 |
| Opus 5 | $0.00012 | $0.03519 |
| Sonnet 5 | $0.00005 | $0.01408 |
| Haiku 4.5 | $0.00002 | $0.00704 |
Grade A, and why
setup scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 504 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are the meta-repo setup command. You convert the current folder into — or upgrade an existing —
meta-repo workspace: an npm/pnpm/yarn/bun root (ROOT_PM, operator's choice) wrapping N
independent git sub-repos, with cross-cutting tooling, parallel worktrees, a ticket queue, a governor
(autonomous ticket loop), and the SessionStart/End/Stop hooks.
The full pattern is documented in the plugin's SKILL.md (read if you need context). All mechanical
file writes are executed by scaffold.sh — you own the interview and judgment; it owns byte-level
writes, deterministically and idempotently.
Locate the plugin
Templates and scaffold.sh live together. Resolve PLUGIN_ROOT in priority order:
${CLAUDE_PLUGIN_ROOT}(plugin run)~/.claude/skills/shiploop/(legacy clone-into-skills)~/.claude/plugins/**cache lookup by plugin name
SCAFFOLD=$PLUGIN_ROOT/scaffold.sh.
Architecture (read once)
Every mechanism script sources ONE config file, scripts/lib/workspace.sh (repo names, dev commands,
ports, GitHub org, worktree base, merge-allowlist). Mechanism scripts are therefore identical across
installs — a bump just refreshes them from templates and only ever (re)generates workspace.sh.
Never hand-edit a mechanism script; put the value in workspace.sh.
Interview doctrine — ONE pause
- Detect first, ask second. Run EVERY detection (mode, sub-repos, ports, dev commands, org, root
PM, worktree base, visibility) AND — in wrap mode —
wrap.sh --preflightBEFORE the first question. Preflight is read-only and surfaces everyREFUSE/NEEDS-CONFIRMitem up front. - One batched interview. Ask everything in a SINGLE
AskUserQuestioncall (4-question limit is enough — see per-mode specs below). Print the detected-defaults table BEFORE the call; detected values aren't questions — override via "Other". - Then run to completion without pausing. Legitimate post-interview stops are ONLY: wrap.sh exit 1 (rolled back), exit 4 (name collision preflight missed), or a NEW exit-5 item preflight could not have seen. Never re-ask something the interview covered.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 504 lines · 24 tokens per session scan A fa6b5a0b6647
setup is a command published in the GitHub repository anshss/shiploop (5 stars, last pushed 29d ago), licensed Apache-2.0. It adds 24 tokens to every session and 7,038 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
work
Execute an approved wish plan — orchestrate subagents per task group with fix loops, validation, and review handoff.
fest-show
Show festival progression (in-progress tasks, roadmap, and dependency view).
upgrade
Comprehensive dependency upgrade with security audit, prioritization, and validation.
review
Review proposed ADRs for lifecycle progression — accept implemented decisions, deprecate stale ones, archive trivial ones.
from-spec
Extract Architecture Decision Records from a completed spec.
types
Debug and fix TypeScript type errors with systematic analysis and expert guidance.