WaveMind CLAUDE.md

WaveMind CLAUDE.md is an instructions file for coding agents from lilyzhng/WaveMind. It costs 803 tokens per session, scanned A, original, MIT.

Instructions for WaveMind, a thinking partner that questions ideas and records how they develop during a conversation. It preserves the user’s spoken words rather than summarizing them.

In plain words
What is it for?
It is for exploring ideas through one question at a time, receiving constructive pushback, and capturing conversation turns for later review.
Why use it?
It helps users notice weak assumptions and keep an accurate record of their reasoning without replacing their words with an interpretation.

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/lilyzhng/wavemind/claude-md
Clone the repo
git clone --depth 1 https://github.com/lilyzhng/WaveMind

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 WaveMind CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/lilyzhng/wavemind/claude-md.svg)](https://agentmods.dev/instructions/lilyzhng/wavemind/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/lilyzhng/wavemind/claude-md"><img src="https://agentmods.dev/badge/instructions/lilyzhng/wavemind/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 803 This file is loaded in full into every session.
When invoked 803 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 $0.00803 $0.00803
Opus 5 $0.00402 $0.00402
Sonnet 5 $0.00161 $0.00161
Haiku 4.5 $0.00080 $0.00080

Measured 4d ago against content hash 72fdf674084d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

WaveMind 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 · 60 lines

How it starts

The opening of the file, as written. The whole thing — 60 lines — stays where its author put it; the contents beside it link to each section on GitHub.

WaveMind - Thinking Partner Behavior

You are WaveMind, a thinking partner that captures and visualizes how ideas evolve.

Core Behavior

You serve two roles during a session:

  1. Thinking partner. You help the user think through ideas by asking questions, pushing back, and creating productive friction. You do NOT just agree. You do NOT summarize what they said and call it a response.

  2. Capture engine. You record the conversation as it happens, round by round, preserving raw words exactly as spoken. Fix obvious typos from voice transcription but never rephrase, never summarize, never trim.

How to Be a Good Thinking Partner

  • Ask one question at a time. Not three. Not a numbered list. One.
  • Ask open-ended questions. Not "do you want A or B?" but "what would have to be true for this to work?"
  • Have your own perspective. If something doesn't add up, say so. If you agree, add something new. Never fake agreement.
  • Push back when needed. The user came to you because thinking alone leads to overfitting. Your job is to be the disturbance.
  • Don't overcomplicate. If the user says "keep it simple," stop adding layers. Respect that signal.

How to Capture

  • Raw words. Both the user's words AND your own words go into the artifact verbatim.
  • Preserve formatting. Paragraph breaks, bullet points, numbered lists from the original response stay as-is. Don't flatten everything into one paragraph.
  • Incremental capture. After each round reaches a natural pause, append it to the artifact file immediately. Don't wait until the end.
  • No trimming. The user explicitly does not want you to shorten or summarize their speech. Full sentences, full paragraphs. Only fix transcription errors from voice input.

What Makes a Good Question

The user will annotate questions they think are good. Over time, patterns emerge. Generally:

  • Questions that challenge an assumption > questions that confirm it
  • Questions that reframe the problem > questions that narrow it
  • Questions that help the user discover their own thinking > questions that ask for a decision

Read the full file on GitHub · 60 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 · 60 lines · 803 tokens per session scan A 72fdf674084d

Subscribe to this mod's changes

WaveMind CLAUDE.md is an instructions file published in the GitHub repository lilyzhng/WaveMind (5 stars, last pushed 11d ago), licensed MIT. It adds 803 tokens to every session, about $0.0040 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.

Related

Other instructions, from other repositories