Research & Writing
Proof
Jalaran Proof reads a mathematical or logical proof one inferential step at a time and marks each move valid, a gap, an error, or unclear — with every verdict yours to override.
Who Proof is for
For students and researchers who want a careful second reader on an argument before a supervisor sees it. The specific value is having gaps named as gaps, because the step you skipped is invisible to you by definition — you believed it.
What Proof does
Proof segments an argument into its inferential moves and assigns each one a verdict: valid, a gap, an error, or unclear. The distinction between a gap and an error is the useful part — a gap is a step that is probably true but unjustified, an error is a step that is wrong, and they call for completely different responses. It is honestly framed as a careful read-through rather than a formal verifier, and every verdict is overridable because the reader is a language model and you are the mathematician.
- Step-by-step verdicts across the whole argument
- Gaps flagged distinctly from errors
- Every verdict user-overridable
- Honest framing — a careful reader, not a verifier
How Proof works
Paste the proof
A mathematical or logical argument, as you have written it.
Read the segmentation
The argument split into individual inferential moves, which is often clarifying on its own.
Check each verdict
Valid, gap, error or unclear per step. Gaps and errors are separated because they need different fixes.
Override what it got wrong
Every verdict is editable. It is a reader, not an authority, and the module is built on that assumption.
What Proof does not do
This is not a computer algebra system and not a formal verifier. It does not check symbolic manipulation, will not catch a subtle algebraic slip reliably, and cannot prove anything. It is a language model reading your argument attentively — genuinely useful for finding unjustified steps, and not remotely a substitute for Lean, Coq or Isabelle if you need actual verification. A clean report is not a certificate of correctness.
Common questions
Does it verify my proof is correct?
No, and it is careful not to claim so. It is a careful read-through that flags steps looking unjustified or wrong. For real verification you need a proof assistant like Lean or Coq.
What is the difference between a gap and an error?
A gap is a step that is probably true but not justified — you skipped the reasoning. An error is a step that is wrong. They need completely different responses, so it never collapses them into one warning.
Why can I override every verdict?
Because the reader is a language model and you are the mathematician. It will sometimes flag a valid step it did not follow, and a tool that would not let you say so would be worse than useless.
Explore the workspace
Jalaran is one workspace of 85 modules. Browse the rest of the arsenal: