TROPT CLAUDE.md

TROPT CLAUDE.md is an instructions file for coding agents from matanbt/TROPT. It costs 3,982 tokens per session, scanned A, original, MIT.

A set of instructions for working on TROPT, a software project that combines models, loss functions, and optimizers for machine-learning experiments.

In plain words
What is it for?
Adding or changing recipes, loss functions, optimizers, model backends, component combinations, documentation, and PyPI releases.
Why use it?
It explains where to look, which project rules to follow, and how to avoid common problems when changing interconnected parts.

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/matanbt/tropt/claude-md
Clone the repo
git clone --depth 1 https://github.com/matanbt/TROPT

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 TROPT CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/matanbt/tropt/claude-md.svg)](https://agentmods.dev/instructions/matanbt/tropt/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/matanbt/tropt/claude-md"><img src="https://agentmods.dev/badge/instructions/matanbt/tropt/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 3,982 This file is loaded in full into every session.
When invoked 3,982 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 $0.03982 $0.03982
Opus 5 $0.01991 $0.01991
Sonnet 5 $0.00796 $0.00796
Haiku 4.5 $0.00398 $0.00398

Measured 4d ago against content hash 46c66e20b3e7, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

TROPT 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 4d 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 · 294 lines

How it starts

The opening of the file, as written. The whole thing — 294 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md

Guidance for Claude Code when working with this repository. For comprehensive design philosophy see DESIGN.md; for step-by-step guides see docs/guides/.

READ FIRST for almost every TROPT task — load skills/tropt/SKILL.md. It's the user-facing companion to this dev-facing file and covers the bulk of what a contributor does day-to-day: adding a recipe, adding a loss, adding an optimizer, adding a model backend, composing custom Model + Loss + Optimizer wirings, swapping components in an existing recipe, debugging cross-cutting pitfalls (mixin mismatches, attention-loss requirements, thinking-model target alignment, black-box vs white-box loss, multi-model OOM, etc.), routing to the right guide / source file, and helping users without a local checkout. Skill loading is not automatic from <repo>/skills/ — the file must be explicitly Read (this is intentional Claude Code behavior, not a bug). Treat the skill as required reading whenever the request touches tropt/recipe_hub/, tropt/loss/, tropt/optimizer/, tropt/model/, or composition patterns. Install is just the supporting first step it also covers.

Tips

  • Prefer concise modifications — minimal changes so edits are easy to review. But if minimal changes create technical debt or unreadable code, prefer clarity.
  • Temporary scripts or markdown you create go under claude_stuff/. Don't make a mess.
  • Avoid over-commenting. Use clear names; only comment where intent isn't obvious.
  • Keep docstrings short. One or two sentences max. Only list args that aren't self-evident from name and type.
  • Docs maintenance: Don't enumerate specific classes/fields/signatures in docs — they go stale. Point to source instead (e.g., "see tropt/loss/ for the full set").

Project Overview

TROPT (Textual Trigger Optimization Toolbox) is a research platform for optimizing discrete text triggers that elicit specific behaviors from NLP models. Primary use cases:

  • Red-teaming: Optimizing triggers toward malicious/undesired model behaviors (LLM jailbreaks)
  • Prompt Tuning: Enhancing desired behaviors through trigger optimization
  • Model Inspection: Crafting adversarial examples and counterfactuals for research

Read the full file on GitHub · 294 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. 4d ago First seen · 294 lines · 3,982 tokens per session scan A 46c66e20b3e7

Subscribe to this mod's changes

TROPT CLAUDE.md is an instructions file published in the GitHub repository matanbt/TROPT (11 stars, last pushed 4d ago), licensed MIT. It adds 3,982 tokens to every session, about $0.0199 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 2026-08-30.

Related

Other instructions, from other repositories

vscode buildNext.instructions.md

Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).

microsoft/vscode · 6,785 tokens

spec-kit AGENTS.md

AGENTS.md instructions for github/spec-kit, covering agents.md, about spec kit and specify, quickstart — add a new integration in 5 steps, integration architecture and integrationmanifest — file tracking.

github/spec-kit · 7,104 tokens

codex AGENTS.md

AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.

openai/codex · 5,182 tokens

langchain AGENTS.md

AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.

langchain-ai/langchain · 4,345 tokens

vscode oss-third-party-notices.instructions.md

Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, applying the notice (cutover) and fallback chain (never fail the build).

microsoft/vscode · 5,001 tokens

next.js AGENTS.md

Instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.

vercel/next.js · 7,296 tokens