Today’s Update: PDF “ask” in MCP Research Friend + MCP sampling support in Chabeau

Today’s Update: PDF “ask” in MCP Research Friend + MCP sampling support in Chabeau

Two workflow improvements that make it easier to research sources and turn them into usable material for Agpedia:

1) MCP Research Friend now supports PDF “ask”

mcp-research-friend can now handle an “ask” request for PDFs: instead of only extracting text, it can ask an LLM to interpret the document and return a targeted result (e.g., summaries, glossaries, key points, Q&A, and other artifacts).

The main advantage is that the entire PDF content doesn’t have to be loaded into your active session/context. It’s closer to tasking a sub-agent with “read this PDF and bring me back what I need,” which helps preserve focus and reduces cost/waste from hauling large documents into the main thread.

2) Chabeau now supports MCP “sampling”

Chabeau has gained support for MCP sampling, the protocol method that allows tools to request LLM help for interpretation and synthesis tasks.

Practically, this means that when mcp-research-friend is active, you can use Chabeau to do things like ask questions about PDFs that come up in a search, and get back answers derived from the document—without turning your working context into a dumping ground for long extracted text.

Why this matters for Agpedia

Together with the existing Agpedia MCP support, these changes make it more feasible to research, synthesize, and prepare Agpedia articles in a single workflow: find sources, interrogate them efficiently, and convert what you learn into draft-ready material.