What is npu-vision-fallback CLAUDE.md?

Project instructions for NPU Vision Fallback, an archived experiment that runs image-processing inference locally and chooses fallback backends when one is unavailable.

Use them when modifying the project, following its architecture constraints, or running its documented build and test commands.

They tell AI coding assistants about the project's local-first, GPU-avoidant design and its dependency and fallback rules.

How Byte-Naut/npu-vision-fallback configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to Byte-Naut/npu-vision-fallback. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/Byte-Naut/npu-vision-fallback/main/CLAUDE.md

Clone the repo

git clone --depth 1 https://github.com/Byte-Naut/npu-vision-fallback

Made for: Claude Code.

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.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/byte-naut/npu-vision-fallback/claude-md/github.svg)](https://agentmods.dev/instructions/byte-naut/npu-vision-fallback/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/byte-naut/npu-vision-fallback/claude-md"><img src="https://agentmods.dev/badge/instructions/byte-naut/npu-vision-fallback/claude-md/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.

Your own site · 80×15

<a href="https://agentmods.dev/instructions/byte-naut/npu-vision-fallback/claude-md"><img src="https://agentmods.dev/badge/instructions/byte-naut/npu-vision-fallback/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 1,799 This file is loaded in full into every session.

When invoked 1,799 The same file — it is already loaded in full.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin 95% copy Near-identical to another mod 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.

Model	Per session	Once invoked
Fable 5.1	$0.01799	$0.01799
Opus 5.5	$0.00720	$0.00720
Sonnet 5.5	$0.00360	$0.00360
Haiku 4.5	$0.00180	$0.00180

Measured 2mo ago against content hash 7712cb42ef85, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

npu-vision-fallback 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 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.

Origin

This is a copy

95% identical to npu-vision-fallback AGENTS.md — 5 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.

CLAUDE.md · 162 lines

How it starts

The opening of the file, as written. The whole thing — 162 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md — Project Constitution

⚠️ Archived (2026-08-01): no longer actively developed. This document remains as the historical engineering record for the experiment.

This file is read by AI coding assistants (Claude Code, Codex, OpenCode) to understand the constraints of this project. Violating these principles requires updating this document and explicit maintainer approval.

It also serves as the day-to-day guide for build/test commands and the high-level architecture. Read it fully before making changes.

Core Principles

Local-first. All inference must run locally. No cloud API calls except in explicit examples/remote_*.py demos.
GPU-avoidant. By default, code paths must NOT wake up discrete GPUs. Prefer NPU > CPU > System OCR > dGPU.
Lightweight core. Core dependencies must not include torch, torchvision, paddlepaddle, or tensorflow. These stay optional. (See Known Deviations for the current ultralytics exception.)
Backend-optional. openvino, ultralytics, rapidocr_onnxruntime, and the Windows winrt-* packages must all be gracefully import-guarded. A missing backend means fallback, not crash.
No screenshot persistence. Never write screenshots to disk by default. Logs must not include OCR text content (privacy).
Coordinate integrity. OCR/detection backends return backend-local coordinates; the Orchestrator is solely responsible for remapping to absolute screen coordinates. Don't double-shift.
Fail loudly, degrade gracefully. An unavailable backend should log a warning and fall through to the next. Never silently return an empty result with ok=True when the real backend is broken.
Test invariants. All tests must pass on Linux CI (using dummy / rapid_ocr). Platform-specific tests are gated by pytest marks: windows_only, macos_only, npu_required.
One tool = one responsibility. Adding a new MCP tool requires updating the README, tests/, and the tool's inputSchema in mcp_server.py.

Read the full file on GitHub · 162 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.

2mo ago First seen · 162 lines · 1,799 tokens per session scan A 7712cb42ef85

Subscribe to this mod's changes

npu-vision-fallback CLAUDE.md is an instructions file published in the GitHub repository Byte-Naut/npu-vision-fallback (1 stars, last pushed 3mo ago), licensed MIT. It adds 1,799 tokens to every session, about $0.0072 per session on Opus 5.5. A static security scan graded it A with 0 findings. It is 95% identical to npu-vision-fallback AGENTS.md, differing in 5 lines, and is treated as a copy.

npu-vision-fallback CLAUDE.md next to 6 instructions from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
npu-vision-fallback CLAUDE.mdByte-Naut/npu-vision-fallback	Instructions file	1	3mo ago	MIT	A	1,799	Claude Code
vscode buildNext.instructions.mdmicrosoft/vscode	Instructions file	195k	yesterday	MIT	A	9,719	GitHub Copilot
next.js AGENTS.mdvercel/next.js	Instructions file	143k	yesterday	MIT	A	7,461	Codex, OpenCode
langchain AGENTS.mdlangchain-ai/langchain	Instructions file	149k	yesterday	MIT	A	4,680	Codex, OpenCode
vscode oss-third-party-notices.instructions.mdmicrosoft/vscode	Instructions file	195k	yesterday	MIT	A	5,516	GitHub Copilot
vscode copilot-instructions.mdmicrosoft/vscode	Instructions file	195k	yesterday	MIT	A	3,093	GitHub Copilot
deepseek-harness AGENTS.mddeepseek-ai/deepseek-harness	Instructions file	274k	2d ago	MIT	A	4,313	Codex, OpenCode

npu-vision-fallback: Instructions file for Claude Code