Experimental Decision Runtime

Inspect the saved decision trail.

Decision Runtime projects an optional view from saved Council artifacts. It exposes recorded claims, risks, dissent, and verification—not hidden model reasoning. The original files and verdict remain authoritative.

What it shows

A structured view of recorded evidence.

Claims and evidence

Collect recorded claims, risks, checks, and source links in one inspectable projection.

Recorded dissent

Keep alternative views beside the selected direction and the evidence that could reopen it.

Local comparison

Compare a Decision Cell projection with a compact export derived from the same saved facts.

Reversible shadow

Inspect, ignore, or remove the projection without changing the source session.

How it fits

The source artifacts stay first.

01

Run Council

Complete the normal review, scoring, dissent, and Chairman synthesis.

02

Save the trail

Keep the session, verdict, evidence links, and recorded checks as the source record.

03

Opt into shadow

Project the additional view only when you explicitly request it.

04

Inspect and compare

Use the projection as an aid; return to the original artifacts for authoritative detail.

Current status

Experimental and non-authoritative.

Original artifacts first

Existing sessions, evidence, scoring, and verdicts remain the source of truth.

Local and opt-in

The shadow stays off by default and outside the authoritative decision path.

No savings claim

No measured token or latency benefit is promised without replay and benchmark evidence.

Source-authoritative If the shadow is missing or unhealthy, ignore it. The original session and verdict remain usable and authoritative.

Use it selectively

Keep the experiment separate from the record.

Existing sessions do not need conversion. The Wiki documents the CLI, safety contract, and recovery path for anyone who chooses to test the shadow view.