Proposal: Add `outcome` field for decision validation tracking

[khalidelborai]

The gap

status tracks where the decision is in its lifecycle (proposed → executed → superseded), but not whether it turned out to be correct. These are different things — a decision can be executed and still be wrong.

Right now if we revert a decision, the only option is to create a new AgDR that supersedes the old one. But that loses the signal: was the original decision bad, or did circumstances change?

What I'm suggesting

An optional outcome field:

outcome: pending       # haven't checked yet (default)
outcome: validated     # confirmed it was the right call
outcome: revisit       # circumstances changed, needs re-evaluation
outcome: reverted      # rolled back, didn't work out

So you'd see things like:

status: executed
outcome: reverted

Which clearly says "we did it, it didn't work."

Why I think this matters

• Teams could actually measure how good their AI-made decisions are
• PR reviewers would know which past decisions were later reversed
• It turns AgDR from a write-once archive into a learning tool

Backward compatible

Totally optional. Existing AgDRs don't need to change. If the field isn't there, it just means nobody's checked yet.

Questions I'm not sure about

• Should there be a ## Outcome Note section in the body for explaining what happened?
• Is revisit meaningfully different from "pending again," or is that confusing?
• Who updates this — the agent, the human, or both?

Would love to hear your thoughts on whether this fits the AgDR philosophy.