Scoring rubric for autolearn tool evaluations. Five criteria, 1-5 scale.

Evaluation Rubric

<Tool Name> -- Evaluation

Skip — excellent open-source voice AI but requires GPU hardware not in our stack

Microsoft VibeVoice -- Evaluation

<p>Automated evaluations of open source tools tested against the current
stack. Each evaluation is produced by the autolearn agent pipeline:
the tool is installed, tested, scored against the
<a href="/wiki/stack-contract.html">stack contract</a>, and documented with real
terminal output.</p>
<h2 id="verdicts">Verdicts</h2>
<ul>
<li><strong>adopt</strong> — Integrate into the stack. Worth the operational cost.</li>
<li><strong>watch</strong> — Promising but not ready or not needed yet. Re-evaluate later.</li>
<li><strong>skip</strong> — Doesn't fit the stack or isn't worth the complexity.</li>
</ul>
<h2 id="evaluations">Evaluations</h2>
<table>
<thead>
<tr>
<th>Tool</th>
<th>Verdict</th>
<th>Score</th>
<th>Date</th>
<th>Summary</th>
</tr>
</thead>
<tbody>
<tr>
<td><a href="/wiki/evaluations/vibevoice.html">Microsoft VibeVoice</a></td>
<td>skip</td>
<td>2.4 / 5.0</td>
<td>2026-04-03</td>
<td>Excellent voice AI but requires GPU hardware not in our stack</td>
</tr>
</tbody>
</table>


Automated evaluations of open source tools tested against the current stack.

Kyle Pericak

Tool Evaluations

Verdicts

Evaluations