Clarive

Clarive provides an interactive loop for creating and refining prompts used with large language models. Users input a description of the desired output; the system generates an initial draft, scores it on criteria such as clarity and completeness, and then iteratively rewrites the prompt while presenting the updated score. The process continues until the score stabilizes, allowing the user to guide the refinement through questions and suggested fixes.

The tool includes a rich‑text editor with formatting options and support for template variables, as well as version history that records each edit and offers word‑level diffs for easy rollback. Workspaces separate prompt libraries by team, and real‑time collaboration shows presence indicators and soft locks, with AI‑assisted conflict resolution when edits overlap.

Additional features let users test a single prompt across multiple model providers—GPT, Claude, Gemini, Llama—in a side‑by‑side matrix, export comparison reports, and import external tool definitions from an MCP server for context‑aware generation. The software is open source under the MIT license and can be run in the cloud or self‑hosted.