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.
npx agentmods add instructions/tpikachu/braincue/claude-mdgit clone --depth 1 https://github.com/tpikachu/BrainCueWhat 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 | $0.01819 | $0.01819 |
| Opus 5 | $0.00910 | $0.00910 |
| Sonnet 5 | $0.00364 | $0.00364 |
| Haiku 4.5 | $0.00182 | $0.00182 |
Grade A, and why
BrainCue 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.
How it starts
The opening of the file, as written. The whole thing — 64 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
What this is
BrainCue — a local-first desktop AI companion for live conversations (Electron + React + TypeScript). It transcribes what it hears in real time (with consent), decides when it can help, and streams grounded cues into a floating, screen-share-invisible overlay ("Cue Card"). It archives each conversation so the next one starts where the last ended, and holds reviewed long-term memory per profile. Data stays on the user's machine; only retrieved context + the current question go to OpenAI (BYO key).
Interviews are one activity, not the product. Interview Copilot and Practice are still fully shipped, but meetings and solo sessions are the daily cases — when you touch a shared path (prompts, retrieval, session defaults), check that it doesn't assume an interview. See docs/00-VISION.md and docs/16-CONTINUITY.md.
One profile at a time. The dashboard is scoped by AppSettings.activeProfileId, chosen in the sidebar switcher and resolved in main (shared/activeProfile.ts) so every window agrees. Profile-scoped surfaces read it — never add another profile <Select> to a page. See docs/19-ACTIVE-PROFILE.md.
The user picks an activity, never a mode. shared/activities.ts is the one catalog: what a call IS (meeting/project/job/subject/personal/game/solo/custom), what a Space is a saved instance of, and which SessionMode the engine derives from it. There used to be two lists answering the same question and they could disagree. Never add a mode picker back to a start surface; add an activity. See docs/18-ACTIVITIES.md.
A Space is where a conversation is kept. archiveSession and extractMemoryCandidates both return 0 when sessions.packId is null — no Space, no summary, no memory. The start modal offers only the Spaces of the chosen activity (spacesFor) and says what will survive the session before it starts; the save prompt offers the choice once more and files the session before either half runs. Exactly one activity refuses to start without one (ActivityConfig.needsSpace, job). See docs/16-CONTINUITY.md §15.
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.
- 2d ago First seen · 64 lines · 1,819 tokens per session scan A 9ec126fffcf2
BrainCue CLAUDE.md is an instructions file published in the GitHub repository tpikachu/BrainCue (5 stars, last pushed 1mo ago), licensed MIT. It adds 1,819 tokens to every session, about $0.0091 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.
Other instructions, from other repositories
UltraRAG AGENTS.md
AGENTS.md instructions for OpenBMB/UltraRAG, covering agents.md, 1) project identity, 2) repository map (what matters most), 3) mental model of the system and 4) two-phase execution lifecycle.
gpt_image_playground AGENTS.md
AGENTS.md instructions for CookSleep/gpt_image_playground, covering agent instructions for gpt-image-playground, 项目概况, 常用命令, 代码风格（强制） and 简单优先.
codecompanion.nvim AGENTS.md
AGENTS.md instructions for olimorris/codecompanion.nvim, covering codecompanion.nvim, commands, code conventions, architecture and general rules.
polar CLAUDE.md
Instructions for jamaljsr/polar, covering claude.md, about polar, essential commands, development and testing individual files.
polar copilot-instructions.md
Instructions for jamaljsr/polar, covering polar - ai coding assistant instructions, core architecture, network orchestration via docker compose, state management (easy-peasy + redux) and visual designer (react flow chart).
ai AGENTS.md
AGENTS.md instructions for vercel/ai, covering agents.md, project overview, repository structure, key directories and core package dependencies.