agentmods.dev

crank-loop

crank-loop is a skill for Claude Code, Codex from luanmorenommaciel/kimi-plugin-cc. It costs 43 tokens per session (511 once invoked), scanned A, original, MIT.

A repeatable process for delegating coding tasks to an agent through written specifications, safety checks, measurement, review, and scoring. A task specification describes the goal, success conditions, allowed files, and a runnable final check.

In plain words
What is it for?
Use it to prepare task files, run delegated implementations, collect timing and cost data, review defects, apply fixes, and score repeated runs.
Why use it?
It reduces uncontrolled delegation by isolating changes, checking tasks before execution, measuring results, and investigating defects before shipping.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

not rated 14repo 3mo ago A scan 43 tokens original MIT

Good fit Use it to prepare task files, run delegated implementations, collect timing and cost data, review defects, apply fixes, and score repeated runs.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/luanmorenommaciel/kimi-plugin-cc/crank-loop
View source ↗ luanmorenommaciel/kimi-plugin-cc
Install

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.

Any agent
npx skills add luanmorenommaciel/kimi-plugin-cc --skill crank-loop
Clone the repo
git clone --depth 1 https://github.com/luanmorenommaciel/kimi-plugin-cc

Made for: Claude Code, Codex.

Wrote 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.

agentmods badge for crank-loop

README.md
[![agentmods](https://agentmods.dev/badge/skills/luanmorenommaciel/kimi-plugin-cc/crank-loop/github.svg)](https://agentmods.dev/skills/luanmorenommaciel/kimi-plugin-cc/crank-loop)
Your own site
<a href="https://agentmods.dev/skills/luanmorenommaciel/kimi-plugin-cc/crank-loop"><img src="https://agentmods.dev/badge/skills/luanmorenommaciel/kimi-plugin-cc/crank-loop/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.

agentmods 80×15 button for crank-loop

Your own site · 80×15
<a href="https://agentmods.dev/skills/luanmorenommaciel/kimi-plugin-cc/crank-loop"><img src="https://agentmods.dev/badge/skills/luanmorenommaciel/kimi-plugin-cc/crank-loop.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 43 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 511 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00043 $0.00511
Opus 5.5 $0.00017 $0.00204
Sonnet 5.5 $0.00009 $0.00102
Haiku 4.5 $0.00004 $0.00051

Measured 2mo ago against content hash 3dd0546b858a, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

crank-loop 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 2mo 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.

plugins/kimi-code/skills/crank-loop/SKILL.md · 38 lines

What it actually says

The Crank Loop

A reproducible process for autonomous, supervised agent cranking. One cycle:

  1. PREP — isolate work (git worktree or clean branch off origin).
  2. SPEC — write a task spec (tasks/T-*.md): goal, success criteria, anti-patterns, touches_paths, and a runnable # Exit Check bash block.
  3. GATE — only delegate specs whose evals are runnable and side-effect-free.
  4. CRANK — execute the spec (e.g. /crank:run), preferring background execution for long tasks.
  5. MEASURE — capture duration, tokens/cost, tool-call mix, and eval pass/fail for every run.
  6. DIAGNOSE — root-cause every defect; adversarially review the result (/crank:challenge).
  7. DESIGN — decide the fixes; plan them (/crank:plan).
  8. APPLY — implement fixes with tests.
  9. SHIP — commit, tag, release.
  10. SCORE — re-crank a fresh task and score against the rubric; repeat until ≥ 9/10.

Scoring rubric

Dimension Weight
Work correctness (task done right) 25%
Isolation (edits land in the right tree) 20%
State preservation (no lost/zombie sessions) 15%
Telemetry (every run measurable) 15%
Scope discipline (no creep) 15%
Auto-commit (work lands committed) 10%

Hard-won rules

  • A hung crank must die: always run with a wall-clock cap AND an idle-output watchdog; a timeout is terminal, never retried.
  • Sessions must be resumable natively (kimi --session <id>) — prompt-text "continue" hacks lose context.
  • Tokens come from the session's real usage records, never from char-count estimates, when available.
  • Read-only roles are a behavior contract (review/explore never write); enforce in the prompt and verify in review.
Changes

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.

  1. 2mo ago First seen · 38 lines · 43 tokens per session scan A 3dd0546b858a

Subscribe to this mod's changes

crank-loop is a skill published in the GitHub repository luanmorenommaciel/kimi-plugin-cc (14 stars, last pushed 3mo ago), licensed MIT. It adds 43 tokens to every session and 511 once invoked, about $0.0002 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 2026-08-31.

Building agentsPlanning
Related

crank-loop compared with skills from other repositories

crank-loop next to 6 skills from other repositories
ModTypeStarsLast pushLicenceScanTokensWorks with
crank-loopluanmorenommaciel/kimi-plugin-ccSkill143mo agoMITA43Claude Code, Codex
imagegenopenai/codexSkill134ktodayApache-2.0A113Codex
playwright-climicrosoft/playwrightSkill100ktodayApache-2.0A19Claude Code
local-ai-agentsmicrosoft/ai-agents-for-beginnersSkill79k3d agoMITA200Claude Code, Codex
browser-automationopenclaw/openclawSkill394ktodayMITA36Claude Code
systematic-debuggingobra/superpowersSkill315k6d agoMITA21Claude Code
next-cache-components-adoptionvercel/next.jsSkill143ktodayMITA95Claude Code