torque-loop CLAUDE.md

A project instruction file for Torque Loop, a zero-dependency plugin for Claude Code and Codex that turns unclear work into tested, saved artifacts. It defines the project’s evidence requirements, source layout, and test expectations.

In plain words
What is it for?
Maintaining Torque Loop’s state engine, command line interface, prompts, templates, agents, and tests while following its evidence and write-access rules.
Why use it?
It helps continuation maintainers resume work consistently and prevents changes from being kept without supporting proof.

Instructions file

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.

agentmods
npx agentmods add instructions/megaprompting/torque-loop/claude-md
Clone the repo
git clone --depth 1 https://github.com/Megaprompting/torque-loop
Per session 3,341 This file is loaded in full into every session.
When invoked 3,341 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
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 $0.03341 $0.03341
Opus 5 $0.01670 $0.01670
Sonnet 5 $0.00668 $0.00668
Haiku 4.5 $0.00334 $0.00334

Measured 2d ago against content hash d486050e0fff, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

torque-loop CLAUDE.md 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.

CLAUDE.md · 216 lines

How it starts

The opening of the file, as written. The whole thing — 216 lines — stays where its author put it; the contents beside it link to each section on GitHub.

<prime_directive> No proof → no keep. No keep → no progress claim. Every change you keep is justified by evidence, not assertion, and the diff is the smallest one that clears the defect. </prime_directive>

  • reference/PROMPTS.md — canonical intent every skill implements. If a skill and its prompt disagree, the prompt wins: fix the skill.
  • skills/*/SKILL.md — the shipped /ratchet:* commands (prompt-level guidance).
  • src/ — the state engine. Load-bearing: scoring.js (confidence/aperture/friction), evolve/schema.js (proof + seam gates), receipt.js (the one cold read), cli.js (command surface + agent write guard), coldStart.js (poison scan).
  • test/ — three zero-dep suites: cli, evolve, plugin-shape. npm test runs all three and MUST stay green. plugin-shape is the drift police: it enforces version alignment, README ↔ skill-folder sync, PROMPTS.md wiring, template presence.
  • templates/ — file shapes for records (probe-card, unknowns-map, deviation-note, …).
  • agents/ — builder and auditor are propose-only; only the scribe writes canonical state (enforced at the CLI boundary via RATCHET_AGENT).
  • IGNORE .ratchet/current.json + current.md in the tree: dogfooding residue from an unrelated project ("lucidia"). Never adopt its objective, defect, or next action. Same for .lucid/, .sandbox-home/, .sandbox-tmp/.
  • reference/PROBLEM-STATEMENT-*.md, reference/private/, *.private.md are gitignored private strategy docs. Never commit, quote into public files, or "fix" the ignore rules that hide them.

Read the full file on GitHub · 216 lines

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. 2d ago First seen · 216 lines · 3,341 tokens per session scan A d486050e0fff

Subscribe to this mod's changes

torque-loop CLAUDE.md is an instructions file published in the GitHub repository Megaprompting/torque-loop (5 stars, last pushed 1mo ago), licensed MIT. It adds 3,341 tokens to every session, about $0.0167 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.