Feat/llm evals - #19
Merged
Merged
Conversation
… with explanation
barlipdev
approved these changes
May 7, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What does this PR do?
Adds model-specialised
MDMA_AUTHORprompt variants for every major LLM family and wires them into the demo's system-prompt picker. Instead of one generic prompt, each vendor/model gets a composition tuned to its known failure modes — derived from running the eval suite (evals/) and iterating until all (or near-all) tests pass.New prompt variants added:
gpt-5,gpt-5-mini,gpt-5-nano,gpt-5.1–gpt-5.5,gpt-5.4-mini/nano,gpt-4.1,gpt-4.1-mini/nanoclaude-opus-4.7,claude-opus-4.6,claude-sonnet-4.6,claude-haiku-4.5gemini-3.1-pro-preview(+ customtools),gemini-3.1-flash-lite-preview,gemini-3-flash-preview,gemini-2.5-pro,gemini-2.5-flash,gemini-2.5-flash-litegrok-4.3,grok-4.20Infrastructure:
evals/select-prompt.mjs— automatic longest-substring routing from model ID to variant file; no config required when adding a new variantpackages/prompt-pack/src/prompts/mdma-author/registry.ts— hand-curated list consumed by the demo picker (AUTHOR_PROMPT_VARIANTS)README.md— prompt matrix showing pass rates across 4 eval suites per variantCloses #
Type of Change
Packages Affected
@mobile-reality/mdma-spec@mobile-reality/mdma-parser@mobile-reality/mdma-runtime@mobile-reality/mdma-attachables-core@mobile-reality/mdma-renderer-react@mobile-reality/mdma-prompt-packHow to Test
Google — Gemini 2.5 Proand send a message asking for a form — verify the response is a valid MDMA document.pnpm run eval:authorin root against any of the new model variants — results should match the README matrix.