Skip to content
Rutger Hemrika
Menu

Writing

#Decision records

10 pieces with this tag.

· written up 20 October 20264 min read

Agents are principals: machine identity as a first-class actor

Every identity source hard-coded the principal type to user, so a caller no human authenticated was recorded as human. Found while building, not by review. Decision record ten.

· written up 20 October 20264 min read

Adopting a vendor agent harness with its dangerous defaults enforced off

A vendor harness reached stable. Four of its default capabilities each break a platform rule. Adopt it, and enforce all four off at two fail-closed points. Decision record nine.

· written up 20 October 20264 min read

Graduated autonomy: delegate by blast radius, keep binding decisions human

Four autonomy classes for a build pipeline that produced decisions faster than a human could ratify them, and the one class that stayed human-only. Decision record five.

· written up 20 October 20264 min read

Break-glass: deny-all by default, and a designed narrow override

When the approval service is down, the override you did not design is the one that gets invented under pressure. The five properties of the one I did. Decision record eight.

· written up 20 October 20264 min read

Content entering the model is untrusted at every entry point

Actions leaving the model were well governed. Content entering it was not. Six entry points, one named kernel control each, and three honest limits. Decision record six.

· written up 20 October 20265 min read

The controls an extension can never disable

A fixed set of kernel controls no extension can replace, and the retrieval filter that was wrongly listed as replaceable. Decision record three.

· written up 20 October 20264 min read

A library the governed code can decline to call is not a control

An in-process governance library can be a useful signal source. It cannot be an enforcement point, because nothing forces the call. Decision record four.

· written up 20 October 20265 min read

Model output is an untrusted candidate until promoted

Every model output starts untrusted and becomes a response, record or memory only through an explicit promotion path. Memory as the worked example. Decision record two.

· written up 20 October 20265 min read

Business rules are enforced outside the model

Prompts configure behaviour; they do not enforce policy. Why executable rules live in a policy service outside the model, and what that costs. Decision record one.

· written up 20 October 20264 min read

Three release gates, and only a human releases

Deterministic checks may block. Statistical evaluation may block, or say inconclusive. Only a named human releases. Why a judge model never gates. Decision record seven.

← Writing