webshift CLAUDE.md

webshift CLAUDE.md is an instructions file for coding agents from x-hannibal/webshift. It costs 1,793 tokens per session, scanned A, original, MIT.

A project instruction file for a Rust web-search library and MCP server, including communication rules, required session setup, project structure, and a reference implementation.

In plain words
What is it for?
Use it when developing the webshift library, its MCP server, or internal tools, especially for code navigation, porting behavior, and repository-specific workflows.
Why use it?
It ensures the coding agent loads the project's tooling and context before making changes, which helps keep development aligned with the existing Rust and Python designs.

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/x-hannibal/webshift/claude-md
Clone the repo
git clone --depth 1 https://github.com/x-hannibal/webshift

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 webshift CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/x-hannibal/webshift/claude-md.svg)](https://agentmods.dev/instructions/x-hannibal/webshift/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/x-hannibal/webshift/claude-md"><img src="https://agentmods.dev/badge/instructions/x-hannibal/webshift/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 1,793 This file is loaded in full into every session.
When invoked 1,793 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 $0.01793 $0.01793
Opus 5 $0.00897 $0.00897
Sonnet 5 $0.00359 $0.00359
Haiku 4.5 $0.00179 $0.00179

Measured 5d ago against content hash 98e09c4ddd29, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

webshift 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 · 177 lines

How it starts

The opening of the file, as written. The whole thing — 177 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.

Communication

Chat in Italian, develop and document in English.

MANDATORY Session Initialization & Tooling

At the start of every session, run Serena's onboarding process to activate the MCP integration and get access to all LSP-powered tools (code navigation, symbol search, references, diagnostics, etc.). Do this before any coding task — it ensures the full toolset is available and the project context is loaded correctly.

Once Serena is active, always prefer its tools over CLI alternatives.

Project

Native Rust port of mcp-webshift (Python). Denoised web search library and MCP server. Two crates in a workspace:

  • webshift — library crate (crates.io). Public API: clean(), fetch(), query(), Config.
  • webshift-mcp — binary crate (crates.io). MCP server over stdio. Installs as mcp-webshift.
  • robot — internal dev tool (not published). bump, test, promote, unpromote, publish.

Reference implementation

The Python source lives in ../mcp-webshift/src/mcp_webshift/. Key files to reference when porting:

  • scraper/cleaner.py — lxml XPath cleaning + text sterilization
  • scraper/fetcher.py — httpx concurrent fetcher, streaming, UA rotation, retry
  • tools/query.py — full search pipeline
  • tools/fetch.py — single page fetch
  • config.py — Pydantic config with env/toml/CLI resolution
  • backends/ — 5 search backend implementations
  • llm/ — OpenAI-compatible client, expander, summarizer
  • utils/reranker.py — BM25 + LLM reranking
  • utils/url.py — sanitize, dedup, binary filter

Stack

  • Rust 2024 edition
  • tokio — async runtime
  • reqwest — HTTP client (streaming)
  • scraper (html5ever) — pure-Rust HTML parsing + CSS selector noise removal
  • serde + toml — config deserialization
  • regex — text sterilization
  • clap — CLI argument parsing
  • rmcp — MCP server SDK (official Anthropic Rust SDK, 1.x stable)
  • serde_json — JSON output

Read the full file on GitHub · 177 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. 5d ago First seen · 177 lines · 1,793 tokens per session scan A 98e09c4ddd29

Subscribe to this mod's changes

webshift CLAUDE.md is an instructions file published in the GitHub repository x-hannibal/webshift (7 stars, last pushed 3mo ago), licensed MIT. It adds 1,793 tokens to every session, about $0.0090 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.