Browse, edit and organize AI instructions, system prompts, and RAG documents for ChatGPT, Claude, Gemini, Cursor, and other LLMs. Export to 26+ agent formats, test in Playground with tool modules, and publish to the community catalog — in one workspace.
Cognitive influence map — see how sections of your prompt depend on each other (prompt dependency graph, instruction flow diagram). The model infers links before you ship to an agent.
Pricing model
The place where you write, check, and ship instructions — without jumping between docs, AI chats, and whiteboards.
Usually
Google Doc, Notion, or a .md file in a folder — everything in one blob of text
What you pay for in aiStructEdit
One document split into clear blocks: role, rules, context, output format — easy to find and edit each part
Usually
Copy into ChatGPT, get a rewrite, paste back — easy to lose context and mix versions
What you pay for in aiStructEdit
Edit in place. The assistant sees the whole document and shows changes before you apply them
Usually
Open n8n or another automation tool: paste the prompt, run the workflow, then jump to chat and manually try to figure out if it worked
What you pay for in aiStructEdit
Built-in Playground: compliance probes, batch QA, interactive dialogues, and tool-runtime demos where the agent calls real actions
Usually
Keep it in your head, in comments, or draw on a whiteboard
What you pay for in aiStructEdit
A map of how sections influence each other — what constrains what
Usually
Re-read everything and hope you catch duplicates and contradictions
What you pay for in aiStructEdit
Automatic scan: repeats, suspicious fragments, token budget, coverage gaps
Usually
Paste into Claude and ask it to reformat for the target system — over and over
What you pay for in aiStructEdit
26+ harness profiles in one menu: Cursor rules, Claude, OpenAI, Gemini, n8n, LangChain, CrewAI — with recent-format history
Usually
Send a file or screenshots; they need their own setup to test
What you pay for in aiStructEdit
Share link: they open a test chat in the browser — no account needed
What the tool is, how it fits your workflow, and how to open the editor — in plain terms.
A browser editor for structured AI content: agent instructions, prompts, and knowledge documents in named sections — with AI assistant, analysis, influence graphs, Playground QA, harness export to 26+ formats, tool-runtime demos, community catalog, and share links.
Unstructured prompts drift, overlap, and contradict each other. Sections keep behavior predictable and make it easier to review coverage, links, and token budget before you ship or index.
A clear document structure, cross-references between sections, AI rewrites with preview, Playground checks, tool modules, influence maps, harness exports for agents and automation, and share links for colleagues.
One-click harness export: Markdown, XML, Cursor rules, Claude and OpenAI message formats, Gemini system instructions, n8n, LangChain, CrewAI, and more — 26 profiles with searchable menu and recent history. Import .xml, .md, or .txt back into the editor.
Test your instruction in chat with compliance probes and batch QA. Attach interactive tool modules — CRM desks, smart home, form wizards, games — where the agent calls [TOOL: …] actions against a live engine panel. Interactive templates add branching dialogues with quick replies.
You can start without paying. Free keeps projects in the browser; paid plans add cloud sync and higher limits. See the comparison table above and Pricing below.
Create a project, split content into sections, refine with AI, run analysis and Playground, then export. Plan limits are listed under Pricing.
Sign in and open the editor. Add your API key in account settings when you want the AI assistant or Playground.
Capabilities
Structure, refine, analyze, test with tools, export to any agent stack, and share — without leaving the editor.
XML-tagged blocks for agent rules, chat prompts, or knowledge articles — reorder, rename, cross-reference, and import existing .xml, .md, or .txt files.
Improve, shorten, translate, suggest structure, red-team weaknesses (Pro+), and apply changes with preview — per section or whole document.
Chat QA with compliance probes and batch runs. Attach tool modules (CRM, IoT, forms, games) where the agent executes real actions. Interactive templates with branching replies.
Prompt linter: duplicates and secrets, token budgets per section, readiness checks, coverage map, and catalog publish validation.
Prompt dependency graph — syntactic @section links plus cognitive maps inferred by the model. Annotate relations back into the document.
26 export profiles for Cursor, Claude, OpenAI, Gemini, n8n, LangChain, CrewAI, and more. Searchable menu, assembly preview, and recent-format history.
Browse and copy community instructions. Publish your own with validation checklist, tags, and AI-generated descriptions. Manage publications from your account.
Organize work by agent, product, or client. Free: browser-local projects. Pro+: cloud sync, multiple documents per project, and migration from local storage.
Send a Playground URL with expiry — colleagues test your instruction in the browser without an account.
Ship & test
Pick a harness profile for your stack, then validate behavior in Playground — with optional tool engines attached to the document.
26 profiles — searchable menu, assembly preview, recent history
Chat QA, batch probes, interactive branching, and tool runtime panels
Tool modules
Interactive templates
Workflow
Split agent instructions, prompts, or knowledge docs into role, rules, context, sources — or your own schema.
Use AI to tighten wording, resolve contradictions, and align tone across related documents.
Map cross-references, check coverage and token budget, run Playground QA where behavior matters.
Export via harness profiles to agents, Cursor, or automation stacks. Share Playground links (Ultra) or publish to the community catalog.
Pricing
Free to start. Paid plans add cloud sync, more providers, and higher limits.
Local browser workspace without cloud sync.
Cloud storage and the full AI toolkit.
No limits for large corpora and custom stacks.
Open the editor with a template — import an existing file, or fork an instruction from the community catalog.