Praxa is now live: theatre research built for people and agents
Praxa has evolved from a theatre RAG prototype into a production-ready, WebMCP-native research assistant.
The latest release introduces three browser-discoverable tools:
ask_theatre— answers theatre questions using retrieved evidence and page-level citations.search_theatre_archive— performs semantic search and returns the most relevant source passages.compare_productions— compares two productions by venue, dates, run details, or another selected focus.
What’s new
- ✅ Three typed WebMCP tools
- ✅ Visible human–agent research workspace
- ✅ Page-level citations and supporting excerpts
- ✅ Structured tool inputs and results
- ✅ Graceful handling of missing evidence
- ✅ Persistent Chroma vector index
- ✅ FastAPI production service
- ✅ CI-gated deployments on Railway
- ✅ Health monitoring and automated tests
For example, Praxa can identify Retrograde by Ryan Calais Cameron, report that it opens at the Apollo Theatre on 8 March 2025, and connect the answer directly to west end.pdf, page 3.
This release demonstrates how WebMCP enables agents to use explicit, trustworthy website capabilities instead of relying on fragile page scraping or simulated clicks. Every tool invocation updates the same visible workspace, allowing the person to inspect the answer, evidence, citations, and execution details.
Try Praxa
Praxa’s goal is simple: enable people and agents to research together while keeping evidence visible and every factual answer traceable to its source.
Log in or sign up for Devpost to join the conversation.