What is google-flow-mcp CLAUDE.md?

Repository instructions for coding agents working on Google Flow MCP, a server that controls Google Flow through a real browser session. Google Flow is a Google Labs tool for generating media with Veo and Nano Banana.

Guiding changes to the server’s browser transport, API discovery, approval and cost checks, and media-generation workflow.

They define safety rules for operations that can spend money, including when to check budgets and how to avoid duplicate generations.

How roshanarnav25-sloth/google-flow-mcp configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to roshanarnav25-sloth/google-flow-mcp. 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/roshanarnav25-sloth/google-flow-mcp/main/CLAUDE.md

Clone the repo

git clone --depth 1 https://github.com/roshanarnav25-sloth/google-flow-mcp

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/roshanarnav25-sloth/google-flow-mcp/claude-md/github.svg)](https://agentmods.dev/instructions/roshanarnav25-sloth/google-flow-mcp/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/roshanarnav25-sloth/google-flow-mcp/claude-md"><img src="https://agentmods.dev/badge/instructions/roshanarnav25-sloth/google-flow-mcp/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/roshanarnav25-sloth/google-flow-mcp/claude-md"><img src="https://agentmods.dev/badge/instructions/roshanarnav25-sloth/google-flow-mcp/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 631 This file is loaded in full into every session.

When invoked 631 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 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.

Model	Per session	Once invoked
Fable 5.1	$0.00631	$0.00631
Opus 5.5	$0.00252	$0.00252
Sonnet 5.5	$0.00126	$0.00126
Haiku 4.5	$0.00063	$0.00063

Measured 2mo ago against content hash 83688163959a, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

google-flow-mcp 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.

CLAUDE.md · 64 lines

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.

Notes for AI coding agents

Guidance for Claude Code and similar agents working in this repo. Humans should read CONTRIBUTING.md instead — it covers the same ground with more context.

What this is

An MCP server that drives Google Flow (labs.google's Veo / Nano Banana tool) through a real browser session. Flow has no public API, but its frontend is a tRPC client — so this speaks that undocumented API first and falls back to DOM clicking only for surfaces with no endpoint.

Read ARCHITECTURE.md before changing the transport layer or the cost gate. It carries the reasoning, not just the shape.

Hard rules

This code spends real money. These are not style preferences.

The cost gate is the point. Never move a budget check to after Approve. Every refusal must happen while rejection is still free.
Never guess a tRPC procedure name. A wrong GET is a 404; a wrong mutation can charge. New endpoints come from flow_discover_api, not from constants.
Never resubmit an in-flight generation. Completion is detected by diffing the media library, precisely so the composer is never touched twice.
Never widen a text match near Approve. checkApprove versus the "Approve, do not ask again" row — see references/ui-playbook.md. A loose match disables the cost gate for the whole session.
This server handles no credentials, ever. It attaches to a browser a human signed into. Re-auth walls stop the run and hand back to the human.
Fail toward a refusal. When wrong about Flow's UI, cost nothing and say which control could not be found.

Before you claim it works

npm run typecheck && npm run lint && npm run format:check && npm test && npm run build

Then confirm the tool surface still loads — a server whose tools fail to register is broken regardless of what compiles:

npm run inspect

Browser-driving code cannot be unit tested. Do not describe it as verified on the basis of a passing build; the README tracks verification status per surface, and it should stay honest.

Read the full file on GitHub · 64 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 · 64 lines · 631 tokens per session scan A 83688163959a

Subscribe to this mod's changes

google-flow-mcp CLAUDE.md is an instructions file published in the GitHub repository roshanarnav25-sloth/google-flow-mcp (16 stars, last pushed 18d ago), licensed MIT. It adds 631 tokens to every session, about $0.0025 per session on Opus 5.5. 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.

Browser automation Creative and media ai-agents browser-automation claude google-flow mcp model-context-protocol

google-flow-mcp CLAUDE.md next to 6 instructions from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
google-flow-mcp CLAUDE.mdroshanarnav25-sloth/google-flow-mcp	Instructions file	16	18d ago	MIT	A	631	Claude Code
vscode buildNext.instructions.mdmicrosoft/vscode	Instructions file	195k	today	MIT	A	9,719	GitHub Copilot
next.js AGENTS.mdvercel/next.js	Instructions file	143k	today	MIT	A	7,461	Codex, OpenCode
vscode oss-third-party-notices.instructions.mdmicrosoft/vscode	Instructions file	195k	today	MIT	A	5,685	GitHub Copilot
langchain AGENTS.mdlangchain-ai/langchain	Instructions file	149k	today	MIT	A	4,680	Codex, OpenCode
vscode copilot-instructions.mdmicrosoft/vscode	Instructions file	195k	today	MIT	A	3,093	GitHub Copilot
vscode model-prompts.instructions.mdmicrosoft/vscode	Instructions file	195k	today	MIT	A	2,022	GitHub Copilot

google-flow-mcp: Instructions file for Claude Code