Inside the Refiner Loop: How OpenPond Learns Without Leaking Context

Refiner reviews completed work after the foreground response has settled. Its job is not to silently rewrite a model or read an account's entire history. It looks for a small, reusable correction that belongs to the Harness—and it can decline to act.

A completed Work result with its Refiner outcome.

The evidence packet is deliberately bounded

Refiner receives the request, user-visible result, normalized tool failures and recovery, short event excerpts, diagnostics, immutable Harness sources, and eligible memory, prompt, Skill, and Agent targets.

It excludes secrets, raw credentials, unrestricted connected-app data, unbounded transcript archives, and unrelated private workspace content. Evidence is constrained by host authorization and byte, time, and cost bounds.

A trace of the Refiner examining a completed task.

The outcome may be no action

After review, Refiner may return no_action, propose a small Harness update, or route the issue to the runtime/product, Taskset, or training owner. A task failure alone is not enough. The system needs evidence that a behavior is reusable, owned by the Harness, within an advertised capability, and suitable for critique and validation.

Changes are inspectable and reversible

An accepted change becomes a new immutable Harness release. That preserves lineage, lets later work use the reviewed version, and provides a rollback point. It is distinct from model-weight training, which needs a Taskset, frozen baseline, learning signal, budget, candidate evaluation, and activation decision.

Continuous-learning controls expose the review path and its state.

The practical promise is modest but powerful: make the smallest evidence-backed change, show why it was allowed, and preserve the ability to undo it. See Refiner evidence for the precise boundary.