Community · E2 · artifact verified
Check rewrites for lost ideas
Lossless Rewrite closes the loop on AI shortening your report: your model rewrites, and Jev checks every protected idea survived - exact wording, meaning, or a reviewed checklist - then helps repair what went missing.
01 · Role in the system
What Jev does here
You select what must stay - a quote kept exactly, meaning preserved through paraphrase, or key ideas extracted from a section into a checklist you review first. Your model rewrites; Jev then checks the result, deciding per protected item whether it survived, and the UI shows which details made it, inspects failures with matching source and output passages, and requests a repair that restores what went missing. The same engine checks an existing draft through the CLI or a browser extension without generating anything, traces each check with explanations, and exports evidence. Use cases named in the README: papers that keep results and caveats, policies that keep conditions and deadlines, emails that keep commitments while changing tone.
02 · Control boundary
Where Jev sits
Protection list to per-item survival judgments after any model's rewrite; deterministic UI maps failures to repair actions and evidence.
Code owns the loop, permissions, thresholds, validation, and side effects. Jev owns only the bounded judgments described above.
03 · Known limits
What this evidence does not prove
- Survival judgments are model opinions with probabilities, not proofs; the repair loop exists because first passes miss things.
- The rewrite itself comes from your model of choice - the record covers the checking and repair layer.
- Browser extension support is labelled experimental in the README.
04 · Attribution
Public sources
This is a Community record: the project was published by a third-party community author.
- dttfrancesco ↗Community · github · public · checked 2026-09-24
- dttfrancesco ↗Community · demo · public · checked 2026-09-24