Research Delegation

What it does

It reads the full PDF you give it, not just the abstract, and follows the paper's own citations outward until the prerequisites behind a claim and the actual pipeline it describes are both clear. Its coverage leans toward specific subtopics in that space: contract-first decomposition, how authority and accountability get handed off between agents, trust and permission handling, and recovery when something in an agentic pipeline goes wrong. Every claim in what it hands back is tied to a source, and its description says plainly that it leaves out any number that wasn't actually in the paper rather than filling a gap with a guess. The end product is a detailed implementation plan, aimed either at a project called Evidence Loom or at whatever project you point it toward instead.

First messages to try

1. Here's a PDF on multi-agent handoff protocols. Read the whole thing, follow the key citations, and tell me what the real prerequisites are before I build anything.
2. Turn this paper's authority-transfer model into a step-by-step implementation plan for a project I'm calling Fieldhand, not Evidence Loom.
3. Where does this paper cite a number you can't independently confirm from the text? Flag those instead of repeating them as fact.

Apps and accounts

Its description mentions a PDF you upload as the input and either Evidence Loom or a project you name as the destination for its plan. It names no outside account or service to connect; the whole job runs on the document and the project you tell it about.

Cautions

• It's built for dense, citation-heavy papers in a specific corner of multi-agent research; a shorter or less formal source may not give it enough to chase.
• An implementation plan built from one paper's framing can still miss practical constraints the paper itself never considered, so treat it as a strong draft, not a finished spec.

Compare with: Research Bot by Arthur MacWatersPodcast Summary Bot by NM