OIA Application Matrix (v0) — feedback & refinement

What this is

A v0 OIA Application Matrix is now published on the committee dashboard:

https://agenticsorg.github.io/community-projects/oia-matrix.html

It maps each application to the OIA 10-layer stack (L0–L9) + 6 spans (security, sovereignty, auditability, energy, identity, provenance), with per-row evidence tiers, inline drill-downs, a gap-closing improvement-prompt generator, and a client-side repo "retest" scan. It is intended to structure the OIA intake discussion — a strawman for committee red-lining, not an official ranking.

This issue is the feedback channel

The page links here. Please use the comments to propose corrections and refinements. Specifically useful:

• Placement corrections — a layer/span assignment that's wrong or overstated (the page flags lean-agentic as one such case where code contradicted the description). Independent re-tagging by a second member is the cheapest way to move rows from desc to code ✓.
• The tagging rubric — a 1-page layer/span rubric with 1–2 anchor examples per layer, so classification is reproducible rather than one person's read.
• Weights — whether/how to add a composite score. Deliberately omitted in v0 because weighting the 12 dimensions is a committee decision, and a single number would launder that judgment.
• New submissions — repos that belong in the matrix. (Recent adds: proffesor-for-testing/agentic-qe, ohdearquant/lionagi.)
• Funnel wiring — how the matrix plugs into intake (metaharness score as the mechanical Stage-2a gate; layer/span footprint as Stage-2b).

Next steps (from the analysis)

1. Adopt the layer-footprint matrix as the standard submission artifact.
2. Wire metaharness score into intake Stage 2a.
3. Draft the 1-page layer/span tagging rubric (anchor examples).
4. Run the first live submission through both and compare the auto report card to a human read.

Start with the published OIA Application Matrix at agenticsorg.github.io/community-projects/oia-matrix.html and review the issue comments for proposed corrections. Compare layer/span placements, the tagging rubric, weighting, new submissions, and funnel wiring against the listed next steps. Done means the committee has agreed on refinements and documented the resulting matrix and intake decisions.