Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add musingfox/cc-plugins/plugin install spiralWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/musingfox/cc-plugins/spiral)<a href="https://agentmods.dev/commands/musingfox/cc-plugins/spiral"><img src="https://agentmods.dev/badge/commands/musingfox/cc-plugins/spiral/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/musingfox/cc-plugins/spiral"><img src="https://agentmods.dev/badge/commands/musingfox/cc-plugins/spiral.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What 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 5.1 | $0.00083 | $0.07220 |
| Opus 5.5 | $0.00033 | $0.02888 |
| Sonnet 5.5 | $0.00017 | $0.01444 |
| Haiku 4.5 | $0.00008 | $0.00722 |
Grade A, and why
spiral 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 10d 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 — 457 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Spiral
You drive the main thread: dispatch Divergence — the one thing here that must not see your hypotheses — dispatch Probes to walk whatever the human leaves open, and a Prototype where a walk can only be settled by building the thing, then converge what they settled into a plan yourself, write what the human reads, and stop. You do not name directions or pick between them.
Each layer lands one plan or milestone, concrete at that layer's grain and no finer. The
next layer diverges from that plan, so the spiral descends — vague question → approach →
milestone → an implementation-sized goal. It stops where the human says it is concrete enough,
and /cf takes the goal from there.
A layer takes as many rounds as it needs. A round is one demand on the human's attention:
every decision that is askable now, asked at once, answered, and folded back in — which is what
makes the next ones askable. concept.md §5 defines the term; this file just runs it.
Concept: ${CLAUDE_PLUGIN_ROOT}/docs/concept.md holds the reasons behind these rules. Read a
section when a step cites it (concept.md §N), not up front. The question is $ARGUMENTS.
mkdir -p .spiral
grep -qxF '.spiral/' .gitignore 2>/dev/null || echo '.spiral/' >> .gitignore
Two counters, and they move independently. The layer L deepens only when the human
says 再挖一層; the round a within a layer increments whenever the layer is not settled yet —
they rejected the menu (都不對), or they answered some decisions and left others blank. Start at
L1-a1. Nothing is ever overwritten — what was asked and what came back are the carry-over that
keeps the next round from repeating it.
1 — Diverge
Agent(subagent_type: "spiral:divergence", model: "opus")with what to widen from plus the carry-over for this layer, verbatim: every direction already rejected and the human's feedback on it, and every decision they have already settled here.
L1-a1→ the question itself.L<N>-a1(N>1) → the previous layer's plan,.spiral/L<N-1>-plan.md.L<N>-a<M>(M>1) → the same source asa1at this layer; there is no plan for this layer yet. What changed is the rejected set and what they settled, not the artifact.- A layer reopened by an ascent (§3, §5) → the same source as
a1at that layer, with the carry-over${CLAUDE_PLUGIN_ROOT}/docs/ascent.mddescribes.
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.
- 10d ago Changed · -11 lines · -9 tokens per session fc7a59120bfa
- 18d ago Changed · +28 lines 8cbbb82d76a1
- 20d ago Changed · +12 lines · +2 tokens per session 5ace6d2423b8
- 1mo ago Changed · +12 lines 96a2b736ff2a
- 1mo ago Changed · +207 lines · +32 tokens per session ce05164ba11c
- 1mo ago First seen · 209 lines · 58 tokens per session scan A 45fbe66bcf07
spiral is a command published in the GitHub repository musingfox/cc-plugins (2 stars, last pushed yesterday), licensed MIT. It adds 83 tokens to every session and 7,220 once invoked, about $0.0003 per session on Opus 5.5. 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 2026-09-05.
Other commands, from other repositories
template
Manage issue templates for streamlined issue creation.
add-note
Add an internal or external note to a ConnectWise PSA ticket.
fest-show
Show festival progression (in-progress tasks, roadmap, and dependency view).
sync-linear
Sync current work with Linear ticket status.
dispatcher
Use when you want the orchestrator to pick the next repo to work on across your whole portfolio — it enumerates candidate repos below the confinement root, resolves free/busy from each repo's session.lock lease, ranks the FREE ones by backlog priority × staleness × readiness, recommends the single most worthwhile one…
workpm
A project-management workflow for coordinating multiple AI workers through five stages. It includes task assignment, shared activity logs, worker replacement, and final checks.