Loading...
Structure system prompts, generate repository CLAUDE.md guidelines, and configure .claude.json tool and command permission rules for Claude Code CLI and Anthropic agent workflows with 100% browser-local privacy.
<system_instructions> <role> Senior Full-Stack Engineer producing clean, modular, and type-safe production code. </role> <project_context> Project: HiMat Enterprise SaaS Stack: Next.js 16 (App Router), React 19, TypeScript, Tailwind CSS </project_context> <coding_standards> - Zero `any` types in TypeScript. - Prefer Server Components over client state. - Maintain high unit test coverage. </coding_standards> <architectural_rules> - Follow domain-driven modular structure in `src/`. - Validate all user inputs with Zod. - Ensure WCAG AA accessibility compliance. </architectural_rules> <execution_guidelines> - Always run static analysis and unit tests before completing tasks. - Keep modifications scoped to necessary files only. - Never output exposed secrets, hardcoded passwords, or unmasked API credentials. </execution_guidelines> </system_instructions>
Choose a starter preset or input your project architecture, tech stack, coding conventions, build commands, and test suites.
Select approved MCP tools (`Bash`, `FileEdit`, `GrepTool`) and define auto-approved terminal commands (`npm test`, `git status`) for frictionless execution.
Export system prompts, copy `CLAUDE.md` to your repository root, and download `.claude.json` directly with zero server transmission.
Zero backend uploads. All prompts, project guidelines, file paths, and secret tokens are processed strictly inside local browser memory.
Generates both human-readable `CLAUDE.md` repository guidelines and machine-readable `.claude.json` configuration files with auto-approved MCP tools and commands.
Configures `approvedTools` and `autoApprovedCommands` (e.g. `npm test`, `git status`, `ls`) to streamline agentic execution without repeated confirmation prompts.
Scans prompt templates and project guidelines for leaked credentials (`sk-`, `ghp_`, Bearer tokens) and auto-redacts them before file export.
Estimates token counts using Anthropic BPE tokenization rules and provides visual context usage metrics against standard 200k context windows.
One-click starter templates for Next.js App Router, Full-Stack Python FastAPI, Monorepo DevOps, and AI Agent MCP Server projects.
HiMat's Claude Code Prompt Generator operates 100% locally inside your client web browser. Neither your system instructions, repository file structures, terminal commands, nor secret API credentials leave your local machine.
It is a developer utility designed to generate, optimize, and sanitize system prompts, `CLAUDE.md` project rule files, and `.claude.json` permission configurations for Anthropic's Claude Code CLI and Claude 3.7 Sonnet agentic workflows.
`CLAUDE.md` is a Markdown file placed at the root of your code repository that provides Claude Code CLI with project architecture guidelines, coding conventions, test commands, and styling preferences.
No. The HiMat Claude Code Prompt Generator operates 100% locally in your web browser. Neither your prompts, file structures, terminal commands, nor API credentials ever leave your device.
`.claude.json` specifies tool and command authorization rules for Claude Code. Defining `approvedTools` (e.g. `Bash`, `GlobTool`, `FileEdit`) and `autoApprovedCommands` allows Claude Code to execute routine commands safely without prompting for manual confirmation.
Yes, 100% free with no account registration required, no daily rate limits, and zero advertisements.
HiMat Technologies builds production AI agent architectures, custom Model Context Protocol (MCP) integrations, automated code refactoring pipelines, and 21-day fixed-scope AI pilots with complete source ownership.
Continue with related utilities, services, and guides from HiMat.