Claude?

NOT REALLYA prototype of the interface is possible, but the durable product depends on difficult systems beyond the UI. Build a safer adjacent workflow instead.

BUILD FIRST

A non-sensitive or read-only version of the core experience
A manual review step wherever correctness or trust matters
A narrow companion tool that removes a real piece of work

LEAVE FOR LATER

Regulated decisions, payments, or irreversible actions
Claims of correctness, compliance, or uptime that the MVP cannot support
Sensitive integrations until security, support, and auditability are ready

RESEARCH NOTES

What this tool actually does.

Checked 2026-09-06

Claude is Anthropic’s conversational AI product with project workspaces, uploaded knowledge, custom instructions, and artifacts. Artifacts are substantial editable documents, code, diagrams, or interactive apps shown beside the conversation; users can iterate, view versions, copy/download, and in supported plans publish or connect them to approved MCP services.

SOURCES

RESEARCHED BUILD PROMPT

Give this to your coding agent.

Build a focused AI project workspace that turns a user request and project files into an editable, shareable artifact. Do not copy Claude's brand, proprietary assets, or UI.

Core workflow
- Let a user create a project with custom instructions, private chat history, and an uploaded knowledge base. Use retrieval to ground answers in project files and show citations or an explicit uncertainty state.
- When the response is a substantial document, code sample, SVG, diagram, or single-page web app, render it in a separate artifact pane with source view, editable text/code, preview, copy/download, and version history.
- Support follow-up edits that target the current artifact, a diff between versions, restore, rename, delete, and a deliberate publish/share action. Ask for confirmation before an artifact uses an external connector.
- Put model/provider, retention, file limits, and personal-versus-shared storage controls in Settings. Keep artifact storage text-only in v1 and show whether shared data is visible to others.

Implementation and scope
- Use React, a Node/Python API, Postgres/SQLite, object storage, a queued document-ingestion worker, an LLM adapter with structured artifact output, and a sandboxed preview iframe with a strict content policy.
- Out of scope: training or reproducing Anthropic models, public artifact discovery, MCP marketplace, autonomous agents, mobile apps, team billing, and arbitrary server-side code execution.
- Seed a project and sample artifact so chat → artifact → edit → version → export works without an external model key.

Include a README with setup, sandboxing, provider configuration, storage/privacy choices, and important limitations.

Visit the official product page