# PromptArch

> Context engineering for AI coding agents. PromptArch generates, lints and version-controls the configuration files agents read: CLAUDE.md, AGENTS.md, Cursor rules, Copilot instructions and memory files. It also builds structured prompts across a set of domains.

Free without an account: the artifact linter (deterministic rules, no sign-up) and one prompt generation per day. Generation beyond that uses credits.

## Start here

- [Home](https://promptarch.ai/): what PromptArch does and who it is for.
- [Lint an agent config](https://promptarch.ai/lint): paste a CLAUDE.md, AGENTS.md, Cursor rule, Copilot instructions or memory file and get a graded report. No account.
- [Try the prompt builder](https://promptarch.ai/try): one free generation, no account.
- [Context engineering guide](https://promptarch.ai/context-engineering): the concepts behind the product.
- [Pricing](https://promptarch.ai/pricing)
- [Full text for LLMs](https://promptarch.ai/llms-full.txt): complete linter rule inventory, size budgets and artifact reference in one file.

## Developer tools

- `npx promptarch lint`: offline linting of agent config files. `npx promptarch init` extracts repo context and writes a full config set.
- MCP server at `https://mcp.promptarch.ai/mcp` (Streamable HTTP). Tools: `promptarch_lint_artifact` (free, no key), `promptarch_list_artifact_types`, `promptarch_generate_artifact` (needs a `pk_` key).

## Artifact generators

One page per agent config format, each explaining the format and its conventions. Index: [https://promptarch.ai/generators](https://promptarch.ai/generators)

- [Cursor Rule](https://promptarch.ai/generators/cursor-rule): Create .mdc rule files for Cursor IDE with frontmatter and activation modes.
- [Cursor Skill](https://promptarch.ai/generators/cursor-skill): Create SKILL.md files for Cursor with YAML frontmatter, triggers, and tool permissions.
- [Windsurf Rule](https://promptarch.ai/generators/windsurf-rule): Create rule files for Windsurf IDE with trigger modes and file patterns.
- [GitHub Copilot Instructions](https://promptarch.ai/generators/copilot-instructions): Create instruction files for GitHub Copilot with repository or path scope.
- [AGENTS.md](https://promptarch.ai/generators/agents-md): Generate AGENTS.md files following the Agent Skills open standard. Vendor-neutral project context recognized by Codex, Jules, Claude Code, Devin, Copilot Coding Agent, Cursor, Windsurf, and Aider.
- [CLAUDE.md](https://promptarch.ai/generators/claude-md): Generate CLAUDE.md project context files for Claude Code. Defines coding standards, architecture, and workflow.
- [Claude Code Rule](https://promptarch.ai/generators/claude-code-rule): Generate a .claude/rules/ file for Claude Code. Path-scoped via a paths: glob list, or always-on when unscoped.
- [Claude Code Skill](https://promptarch.ai/generators/claude-code-skill): Create skill files for Claude Code with YAML frontmatter, triggers, and tool permissions.
- [Claude Project Instructions](https://promptarch.ai/generators/claude-project): Create project instructions for Claude.ai Projects with context, rules, and workflow preferences.
- [Outcomes Rubric](https://promptarch.ai/generators/outcomes-rubric): Generate evaluation rubrics for Anthropic's Outcomes-style self-grading loop. A separate evaluator agent scores task-agent output against your criteria.
- [Subagent Spec](https://promptarch.ai/generators/subagent-spec): Define a lead orchestrator or specialist subagent for Anthropic's Multi-Agent Orchestration. Agents coordinate via shared filesystem, not message passing.
- [OpenAI Custom GPT](https://promptarch.ai/generators/custom-gpt): Create system instructions for OpenAI Custom GPTs with capabilities, persona, and behavior rules.
- [GPT System Prompt (API)](https://promptarch.ai/generators/gpt-system-prompt): Create system prompts optimized for OpenAI's API with model-specific tuning and function calling support.
- [Context Pack (all agents)](https://promptarch.ai/generators/context-pack): Write once, run every agent. Generate one canonical project context and export AGENTS.md, CLAUDE.md, Cursor rules, and Copilot instructions together.
- [MCP Tool Description](https://promptarch.ai/generators/mcp-tool-description): Generate JSON tool definitions for the Model Context Protocol (MCP) with schemas and examples.
- [Agent Task Specification](https://promptarch.ai/generators/agent-task-spec): Define structured task specifications for multi-agent systems with objectives, contracts, and handoff rules.
- [Agent Memory File](https://promptarch.ai/generators/memory-file): Define how a persistent agent keeps memory: a MEMORY.md policy + seed index with one-lesson-per-file conventions, update triggers, and retention rules.
- [General System Prompt](https://promptarch.ai/generators/system-prompt-general): Create model-agnostic system prompts optimized for any LLM application: chatbots, RAG pipelines, agents, and more.

## Prompt domains

- [Autonomous Agents](https://promptarch.ai/templates/autonomous-agents)
- [Product Mockup](https://promptarch.ai/templates/mockup-builder)
- [Creative Writing](https://promptarch.ai/templates/creative-writing)
- [Software Development](https://promptarch.ai/templates/software-dev)
- [Business Communications](https://promptarch.ai/templates/business-comms)
- [Guides & Documentation](https://promptarch.ai/templates/guides-docs)
- [Image Generation](https://promptarch.ai/templates/image-gen)
- [Data & Analysis](https://promptarch.ai/templates/data-analysis)
- [Marketing Campaigns](https://promptarch.ai/templates/marketing-campaigns)
- [SEO & Content Strategy](https://promptarch.ai/templates/seo-content)
- [Education & Learning](https://promptarch.ai/templates/education-learning)

## Comparisons

- [PromptArch vs PromptBase (Prompt marketplace)](https://promptarch.ai/compare/promptbase)
- [PromptArch vs AIPRM (ChatGPT prompt library)](https://promptarch.ai/compare/aiprm)
- [PromptArch vs PromptPerfect (Prompt optimizer)](https://promptarch.ai/compare/promptperfect)

## Config gallery

Real-world agent configs by stack, each with a lint grade. Index: [https://promptarch.ai/gallery](https://promptarch.ai/gallery)

- [Next.js + Supabase](https://promptarch.ai/gallery/nextjs-supabase)
- [FastAPI](https://promptarch.ai/gallery/fastapi)
- [TypeScript Monorepo](https://promptarch.ai/gallery/turborepo)
- [React Native + Expo](https://promptarch.ai/gallery/react-native-expo)
- [Go Service](https://promptarch.ai/gallery/go-service)
- [Cloudflare Workers](https://promptarch.ai/gallery/cloudflare-workers)
- [Django](https://promptarch.ai/gallery/django)

## Writing

- [How to Write a CLAUDE.md That Claude Code Actually Follows (2026 Guide)](https://promptarch.ai/blog/how-to-write-a-claude-md): How to write a CLAUDE.md for Claude Code: what to include, what to cut, a full example file, the 200-line target, and free tools to generate and lint yours.
- [How to Write a Cursor Rule the Model Actually Follows: .mdc Anatomy, Rule Types, and What to Leave Out](https://promptarch.ai/blog/how-to-write-a-cursor-rule): How to write a Cursor Rule step by step: .mdc frontmatter, the four rule types, glob scoping, research-backed anti-patterns, plus a free generator and linter.
- [How to Write a Claude Code Skill: SKILL.md, Frontmatter, and Descriptions That Actually Trigger](https://promptarch.ai/blog/how-to-write-a-claude-code-skill): How to write a Claude Code Skill: the SKILL.md format, YAML frontmatter, descriptions that trigger reliably, and the mistakes that stop skills from firing.
- [How to Write a Cursor Skill: SKILL.md Anatomy, Trigger Descriptions, and What Belongs in a Skill vs a Rule](https://promptarch.ai/blog/how-to-write-a-cursor-skill): Learn how to write a Cursor Skill: SKILL.md anatomy, frontmatter, descriptions the agent actually triggers on, and skill-vs-rule scoping, with a full example.
- [Prompt Engineering in July 2026: The Prompt-Stripping Era](https://promptarch.ai/blog/prompt-engineering-july-2026-the-prompt-stripping-era): Within nine days, both Anthropic (Claude Fable 5) and OpenAI (GPT-5.6) shipped official guidance that says the same thing: subtract from your prompts, don't add. Plus what changed for prompt engineers, and what we shipped into PromptArch.
- [The PromptArch CLI & MCP Server: A Complete Guide](https://promptarch.ai/blog/promptarch-cli-and-mcp-server-guide): Lint and generate CLAUDE.md, AGENTS.md, Cursor rules, and Copilot instructions from your terminal with the promptarch CLI, or let your AI agent do it over MCP. Setup, commands, API keys, and CI usage.
- [Prompt Engineering in June 2026: Honest Agents and the Caching Cost Crunch](https://promptarch.ai/blog/prompt-engineering-june-2026-honest-agents-and-the-caching-cost-crunch): Two big shifts in late May and early June 2026: Claude Opus 4.8 optimizes for honesty over benchmarks, and Anthropic's June 15 billing change turns prompt-cache hit-rate into a line item. Plus harness engineering, MCP security, and two frontier models on the horizon.
- [Prompt Engineering in May 2026: The Model Generation Shift](https://promptarch.ai/blog/prompt-engineering-may-2026-model-generation-shift): Three major model releases in three weeks: Claude Opus 4.7, GPT-5.5, and Gemini 3. What changed for prompt engineering and what to do about it.
- [Chain of Thought Prompting: A Practical Guide for 2026](https://promptarch.ai/blog/chain-of-thought-prompting-guide): Learn what chain of thought prompting actually does, how to structure reasoning traces, model-specific CoT behaviors, and how to score reasoning quality before shipping.
- [Context Engineering: Why Prompt Engineering Is No Longer Enough](https://promptarch.ai/blog/context-engineering-the-new-paradigm): The field has evolved from crafting clever prompts to designing entire information systems for AI. Here's what context engineering means, why it matters, and how to start practicing it.
- [AI System Prompt Best Practices 2026 - Agents, Models & Examples](https://promptarch.ai/blog/ai-system-prompt-best-practices-2026): The 2026 playbook for AI system prompts in production. Covers prompt anatomy, agent tool-use rules, Claude/GPT/Gemini conventions, and worked examples.
- [PromptPerfect Shutting Down? Migrate to PromptArch in Minutes](https://promptarch.ai/blog/promptperfect-shutting-down-alternative): PromptPerfect shuts down Sept 2026. Switch to PromptArch for guided prompt builders, quality scoring, and Claude/GPT/Gemini output. Your first prompt is free.
- [PromptArch v2: Introducing the Context Engineering Studio](https://promptarch.ai/blog/promptarch-v2-context-engineering-studio): PromptArch v2 launches with the Context Engineering Studio: 12 artifact types for building system prompts, cursor rules, CLAUDE.md files, MCP tool descriptions, and more.
- [Prompt Engineering March Monthly Briefing](https://promptarch.ai/blog/prompt-engineering-march-2026-monthly-briefing): A comprehensive summary of the biggest AI and prompt engineering developments from late March 2026: Mistral Small 4, Gemini 3.1 Pro's distinct prompting rules, Claude Code auto mode, GPT-5.4 mini, and the formalization of context engineering.
- [6 Prompt Engineering Updates You Need to Know This Week](https://promptarch.ai/blog/prompt-engineering-weekly-highlights-2026-w12): A quick recap of the biggest AI and prompt engineering developments from the week of March 16–22, 2026, from GPT-5.4 to the rise of context engineering.
- [All posts](https://promptarch.ai/blog) · [RSS](https://promptarch.ai/feed.xml)

## Optional

- [Research notes](https://promptarch.ai/research)
- [Terms](https://promptarch.ai/terms) · [Privacy](https://promptarch.ai/privacy)

Not crawlable: authenticated product surfaces (dashboard, library, studio, settings) are noindex and require an account.