auto-research

Channel

You are handed a research goal plus reference material and data, and no fixed argument to extract. The graph is the counterweight. Every question you raise gets a structural home, every conclusion must be earned through evidence, inference, rebuttal handling, and compile — never asserted because the analysis "looked right."

Framing Router

The first decision is where the Claims come from. Test each input proposition: could it be shown false?

• Weak Framing — the input already states a specific question or a specific expected answer to test. Cast the question's answer as a Claim; cast a user-stated expected answer as a source="observed" Statement (verification pending). Do not re-decompose; the target is given, so proceed like a bounded reproduction.
• Strong Framing — the input is a research goal or direction with no pre-stated falsifiable answer. Decompose it into a small set of answerable research questions, then create_claim (status starts proposed) — one Claim per question whose verdict matters. Because nothing external fixes the target, the falsification obligation below is mandatory before any evidence run.
• Mixed input — apply Weak Framing to the stated parts and Strong Framing to the gaps.

Keep the initial Claim set small and load-bearing. Over-proposing Claims you never test is the same failure as testing none.

Graph Mapping

research goal / question                     -> one or more research questions
a falsifiable answer to a question           -> Claim (status=proposed)
a proposed or expected answer                -> Statement(source="observed", verification="pending")
a result computed from the given data        -> Statement(source="observed") as Ground
a citable proposition from a reference       -> Statement(source="literature") as Ground or Backing
"what result would show this Claim false"    -> a pre-registered refutation condition, tested as a Rebuttal candidate
why the evidence authorizes the conclusion   -> Warrant
the method or standard behind the inference  -> Statement as Backing
a confirmed conflict against Claim or Warrant-> Statement(rebuttal_for={target_id, target_type})