@mobile-reality/mdma-prompt-pack

0.4.0

4a04d6f: Make the mobile-reality/mdma-il author variant DSL-aware. The MDMA-IL model reads an MDMA-IL DSL intent, so its system prompt must describe the DSL grammar; the previous variant had none. getAuthorPromptVariant('mobile-reality/mdma-il') now returns the full authoring prompt — DSL input grammar, authoring rules, and worked form/table/chart examples — as the single source of truth (previously duplicated in the eval harness). The registry label/description are unchanged.

5bb8529: Split validator into per-block validate() and multi-message validateConversation(); make form.onSubmit required and rewrite action-label fields as opaque labels (drop the action-references rule); add many model-specific fixer/author/agent-tool prompt variants (gpt-5.x family, Claude opus/sonnet/haiku, Gemini 2.5/3, Grok), promote the conversation-judge prompt out of mdma-fixer/ and rename its export to MDMA_CONVERSATION_JUDGE.
Updated dependencies [5bb8529]
- @mobile-reality/mdma-spec@0.3.0

d23b52c: Added AGENT_TOOL_PROMPT_VARIANTS registry with model-specific agent tool prompt variants for Anthropic, OpenAI, Google, and xAI models; changed sensitive input toggle to use 👁/🔒 icons

4b595c8: Add model-specialised MDMA_AUTHOR prompt variants for OpenAI, Anthropic, Google, and xAI. New getAuthorPromptVariant(modelId) export for selecting the optimized prompt variant per model.

9ba720a: Add file field type to forms. Forms can now declare file upload inputs, with a default file input UI in the renderer (overridable via ElementOverridesContext), schema defaults in the validator, and authoring guidance in the prompt pack.
Updated dependencies [9ba720a]
- @mobile-reality/mdma-spec@0.2.1