# AI use cases for Veri Ink (znxzsy/VeriInk) Repo: [znxzsy/VeriInk](https://github.com/znxzsy/VeriInk) | Use cases: 3 | Page: https://dedupedai.com/use-cases/znxzsy-veriink/ VeriInk is a Python tool that renders deterministic synthetic handwriting worksheets from model failures, recording seeds, labels, ink bounds, and file hashes for audit. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Rebuild the audit set and verify every hash: https://dedupedai.com/use-cases/znxzsy-veriink/rebuild-audit-set-and-verify-hashes/ - Turn a model failure into a reproducible handwriting batch: https://dedupedai.com/use-cases/znxzsy-veriink/turn-model-failure-into-handwriting-batch/ - Regression-test an OCR model on six synthetic hard cases: https://dedupedai.com/use-cases/znxzsy-veriink/regression-test-ocr-on-synthetic-hard-cases/ ## How to rebuild the audit set and verify every hash with Veri Ink Repo: [znxzsy/VeriInk](https://github.com/znxzsy/VeriInk) | Difficulty: Intermediate | Time: 15 minutes | Basis: From the README Regenerate the 16-page demo and check that all images match their recorded SHA256 values. **Who it is for:** An ML engineer or auditor who needs proof that a handwriting dataset build is deterministic and unmodified. **The problem:** Training and regression data is hard to trust when you cannot rebuild it or confirm the files were not changed. **The outcome:** A local copy of the 16-page, 80-region audit set, with a passing pytest run that confirms determinism and input checks. ### Steps 1. Run `git clone https://github.com/znxzsy/VeriInk.git` and then `cd VeriInk`. 2. Run `python -m venv .venv` and then `source .venv/bin/activate`. 3. Run `pip install -e ".[dev]"`. 4. Run `python scripts/generate_demo.py`, which rebuilds the examples and recomputes every hash before it exits. 5. Run `pytest -q` and confirm that all tests pass. ### Prompt to paste into your agent ```text Clone https://github.com/znxzsy/VeriInk, create a virtual environment, install it with pip install -e ".[dev]", run python scripts/generate_demo.py, then run pytest -q. Report the number of pages and answer regions produced and whether pytest passed. Verify that examples/gallery/manifest.json lists a SHA256 for each image. ``` ### Caveat The README describes the demo build and tests, but it does not say how long the run takes or what hardware it needs. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/znxzsy-veriink/rebuild-audit-set-and-verify-hashes/ ## How to turn a model failure into a reproducible handwriting batch with Veri Ink Repo: [znxzsy/VeriInk](https://github.com/znxzsy/VeriInk) | Difficulty: Advanced | Time: 30 minutes | Basis: From the README Trace a failed case to its seed and generation record, then regenerate a matching batch. **Who it is for:** An ML engineer whose OCR or multimodal grader just failed on a specific handwriting pattern and needs more data like it. **The problem:** A failure is hard to reproduce or expand into training data when the generation details were never recorded. **The outcome:** A clear view of the seed, labels, bounds, and hash for the failing case, plus a regenerated batch built from the same rules. ### Steps 1. Run `git clone https://github.com/znxzsy/VeriInk.git` and then `cd VeriInk`. 2. Read the architecture and data boundary notes in the docs/ folder. 3. Open src/veriink/ and locate the deterministic renderer and the validation code. 4. Run `python scripts/generate_demo.py` to rebuild the gallery and its manifests. 5. Open examples/gallery/manifest.json and check the seed, labels, bounds, and SHA256 for the record that matches your failure pattern. ### Prompt to paste into your agent ```text In the VeriInk repo, read docs/ and src/veriink/, then open examples/gallery/manifest.json. For the failure pattern , list the matching records with their seed, labels, bounds, and SHA256. Verify that each listed seed, when used for rendering, is covered by the existing validation checks. ``` ### Caveat The README does not document how to write a task specification from a failure, so you will need to read the source to map your own failure into the renderer. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/znxzsy-veriink/turn-model-failure-into-handwriting-batch/ ## How to regression-test an OCR model on six synthetic hard cases with Veri Ink Repo: [znxzsy/VeriInk](https://github.com/znxzsy/VeriInk) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Check a transcription model on crossed-out, decimal, and mixed-operation handwriting with public hashes. **Who it is for:** An evaluator who maintains an OCR or multimodal grading model and needs a fixed regression set that contains no real student images. **The problem:** Public handwriting benchmarks may not cover crossed-out revisions and arithmetic patterns, and real student images carry privacy limits. **The outcome:** A repeatable regression check where the model is scored on the visible answer, with each image confirmed against its published hash. ### Steps 1. Run `git clone https://github.com/znxzsy/VeriInk.git` and then `cd VeriInk`. 2. Open examples/hardcases/manifest.json to list the six synthetic images and their public hashes. 3. Run `sha256sum examples/hardcases/*` and compare each output with the matching hash in the manifest. (inferred, check the README) 4. Run your OCR or multimodal model on the six images and record its transcriptions. 5. Compare each transcription with the visible answer, not a corrected answer, since wrong answers stay wrong by design. ### Prompt to paste into your agent ```text Clone https://github.com/znxzsy/VeriInk, read examples/hardcases/manifest.json, and verify the SHA256 of each synthetic image. Run on the six images and report, for each image, whether the transcription matches the visible answer. Verify that every hash matches before you score the model. ``` ### Caveat The six hard cases cover only the listed arithmetic and revision patterns, so they do not stand in for a full handwriting benchmark. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/znxzsy-veriink/regression-test-ocr-on-synthetic-hard-cases/