Feedback
If you came here from the NAFEMS Americas 2026 talk, thank you. Three things to consider, in order of how much they help.
1. Run UofA on your own evidence
Section titled “1. Run UofA on your own evidence”This is the most useful thing you can do. The CLI is fully local. No data leaves your environment.
pip install 'uofa[excel]'
# Point uofa extract at a folder of your existing evidence# (PDFs, Word docs, validation reports, solver configs, acceptance criteria)uofa extract path/to/your/evidence-folder/ -o my-assessment.xlsx
# Open my-assessment.xlsx and review what was extracted# (this is where you catch what the LLM missed)
# Generate a signing key onceuofa keygen keys/my-assessment
# Import the reviewed workbook to a signed UofA package, validate, and run the rule engineuofa import my-assessment.xlsx --sign --key keys/my-assessment.key --checkThe uofa import --check output is a personalized weakener report. It tells you which credibility gaps the rule engine surfaces in your evidence package, with severity and affected node IDs. Critical and High firings are the ones to address before submission. The full pipeline runs locally — no data leaves your environment.
If your existing documentation is in Word and Excel rather than JSON-LD, the Excel on-ramp takes you from a filled workbook to a signed JSON-LD package in one command. If your existing documentation is in source PDFs, the live demo walkthrough shows how uofa extract ingests an evidence folder and produces the workbook for you.
2. Tell me what gaps you see
Section titled “2. Tell me what gaps you see”The most useful kind of feedback is concrete. “Pattern W-EP-04 fires on a case where I would not flag the evidence as weak” is more actionable than a thumbs-up or a thumbs-down. The Phase 2.5 catalog refinement that produced the current pattern set was driven entirely by exactly that kind of feedback.
Two channels:
| Channel | Best for |
|---|---|
| GitHub Discussions | Public technical discussion, pattern requests, integration questions |
| Email (support@uofa.net) | Confidential or company-specific feedback |
3. Tell me what would make UofA useful for your pipeline
Section titled “3. Tell me what would make UofA useful for your pipeline”Two questions I am actively trying to answer.
What would it take for UofA packages to flow through your existing PLM, SPDM, or CAE pipeline? What format conversion, what API surface, what metadata bridge?
What domain pack would unlock the most value for you? The current catalog ships
vv40(FDA medical device, ASME V&V 40) andnasa-7009b(NASA aerospace). DO-178C, automotive ISO 26262, ISO 42001 AI Management Systems, and an SBKF aerospace shell-buckling pack are on the post-defense roadmap. Tell me which one first.
Or fill out the structured feedback form, which captures these two questions plus a few about your domain, use case, and submission timeline. Takes about 3 minutes.
Email, a GitHub Discussion thread, or the structured form are all read. I respond to every one.
What you will see in the v0.6+ catalog
Section titled “What you will see in the v0.6+ catalog”Six Tier 1 candidate patterns are gated on Phase 3 expert validation. They are not pre-shipped. They are: W-EV-01 (stale validation data), W-EV-02 (inadequate validation metric), W-REQ-01 (ambiguous acceptance criterion), W-CX-01 (configuration divergence), W-AR-06 (eliminative argumentation absent), W-AR-07 (sustained defeater without residual-risk justification). Three or more confirmed by the expert panel ships the next minor.
If any of these resonate with patterns you have seen in real submissions, that is exactly the practitioner signal the Phase 3 protocol is designed to capture. Drop a note.
This is an open-source academic project. The code and schema are both Apache-2.0. There is no sales contact and no funnel. The single best way to support the work is to run it on your own evidence and tell me what you find.