helm CLAUDE.md

helm CLAUDE.md is an instructions file for coding agents from tylergibbs1/helm. It costs 419 tokens per session, scanned A, original, MIT.

Repository instructions for Helm, a browser-automation server that lets software interact with web pages using human-readable labels. They explain its Bun-based setup, structure, configuration, and tests.

In plain words
What is it for?
Use them to run, configure, type-check, and test Helm, including its browser navigation, observation, interaction, and session tools.
Why use it?
They keep development consistent with the project's chosen tools and design decisions. This avoids using unsupported commands or interacting with pages through brittle technical selectors.

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/tylergibbs1/helm/claude-md
Clone the repo
git clone --depth 1 https://github.com/tylergibbs1/helm

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 helm CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/tylergibbs1/helm/claude-md.svg)](https://agentmods.dev/instructions/tylergibbs1/helm/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/tylergibbs1/helm/claude-md"><img src="https://agentmods.dev/badge/instructions/tylergibbs1/helm/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 419 This file is loaded in full into every session.
When invoked 419 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 5.1 $0.00419 $0.00419
Opus 5 $0.00210 $0.00210
Sonnet 5 $0.00084 $0.00084
Haiku 4.5 $0.00042 $0.00042

Measured 5d ago against content hash fede8d8d0223, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-05, from the pricing page.

Security

Grade A, and why

helm 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 5d 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 · 50 lines

What it actually says

Helm — Semantic Browser Automation MCP Server

Default to using Bun instead of Node.js.

  • Use bun <file> instead of node <file> or ts-node <file>
  • Use bun test instead of jest or vitest
  • Use bun install instead of npm install
  • Use bun run <script> instead of npm run <script>
  • bun:sqlite for SQLite. Don't use better-sqlite3.

Project Structure

  • src/server.ts — MCP server entrypoint (stdio transport)
  • src/tools/ — Tool definitions (navigation, observation, interaction, composite, session)
  • src/core/ — Internal modules (browser, resolver, observer, som, memory, recovery)
  • src/types.ts — Shared type definitions
  • tests/ — Unit and e2e tests using bun test + Playwright

Running

bun run src/server.ts        # Start MCP server (stdio)
bun test                     # Run tests
bunx tsc --noEmit            # Typecheck

MCP Config

{
  "mcpServers": {
    "helm": {
      "type": "stdio",
      "command": "bun",
      "args": ["run", "/path/to/src/server.ts"]
    }
  }
}

Key Design Decisions

  • Tools are semantic (click by label, not selector). The resolver handles label→element mapping.
  • observe(task) returns filtered elements, not full a11y tree. Keeps token cost low.
  • Set-of-Mark (SoM) fallback: screenshot_som() + click_mark(id) for sites with poor ARIA.
  • Site memory: SQLite in data/memory.sqlite, keyed by domain. Records successful actions.
  • Error recovery: auto-retry with backoff, auto-dismiss cookie/modal overlays.
  • Zod schemas define tool inputs. MCP SDK wires them to JSON-RPC.
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. 5d ago First seen · 50 lines · 419 tokens per session scan A fede8d8d0223

Subscribe to this mod's changes

helm CLAUDE.md is an instructions file published in the GitHub repository tylergibbs1/helm (1 stars, last pushed 6mo ago), licensed MIT. It adds 419 tokens to every session, about $0.0021 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.