Proposed specificationSpecification onlyH3

On-demand adaptation target selection

Choose harness versus weight adaptation.

JOIN POINTembodied_adaptation_decisionRECORD STATUSCandidate · v0.1.0

UML class view

The proposed contract drawn as participant roles, received calls, and numbered dependencies.

Object interactions

The same roles and calls as a class collaboration and a numbered message sequence.

Object interaction diagram for On-demand adaptation target selection, showing role collaboration and a numbered message sequence — FIG. E08 / OBJECT INTERACTIONS
Open full SVG Download PDF

Pointcut

Where the mechanism applies. All four dimensions belong to the proposed match, not just its vocabulary.

Context: An embodied agent encounters a task or distribution shift.
Trigger: An adaptation controller evaluates task performance and cost.
Structure: Adaptation target policy, harness and weight candidates, and evaluation feedback.
Effect: The controller selects a declared adaptation target and records why.

Protected contract

The obligation that clients, operators, and auditors should be able to rely on.

Requires: Task state, candidate adaptation targets, and cost or risk are observable.
Ensures: The selected target is compatible with the current execution environment.
Invariant: Adaptation does not silently combine targets outside the declared policy.
Violation: Unsupported target or uncertain evidence falls back to a safe target or no adaptation.

Negative control

A paper citation or generic fine-tuning script is not an E08 instance.

Variation & lifecycle

The hot spot is admitted only where the protected semantics survive.

Permitted variation

Harness adaptation, weight adaptation, recursive adaptation, and embodied task domain.

Test obligation

Test target choice, unsupported change, reversibility, and transfer.

Evolution question

Is there an official implementation with target-selection provenance?

Prompt + code design

An illustrative prompt and a small design sketch for the proposed contract. Names and APIs are invented; this is not a repository excerpt.

PROMPT SKETCH

Observe the task state and the candidate adaptation targets. Choose a target that is compatible with the current environment, and record why that target was selected.

CODE DESIGN / PYTHON

def adapt(state, candidates, policy):
    viable = [
        candidate for candidate in candidates
        if candidate.compatible(state.environment)
    ]
    if not viable:
        return NoAdaptation("no compatible target")
    chosen = policy.choose(viable, state)
    return Adaptation(
        target=chosen.id,
        kind=chosen.kind,
        evidence=chosen.compatibility_record,
    )

Illustrative design sketch. Harness and weight adaptation are different targets with different evidence. This example is a teaching aid, not implementation evidence or a conformance claim.

Detection & matching

The frozen rule that retrieves this record, the canonical source exemplar, and the control that should stay unmatched.

Join point: embodied_adaptation_decision
Source locator: Wang et al.
Lexical cue: (?i)\b(?:adaptation[_-]?target|weight[_-]?adaptation|on[_-]?demand[_-]?adaptation)\bFrozen expression, v0.1.0. A match is a retrieval cue, not a confirmed implementation.
Aliases: adapttarget, targetselection, domainadapt, adaptationtarget
Canonical exemplar: None. This record is anchored only in its source specification.
Negative control: A paper citation or generic fine-tuning script is not an E08 instance.

Scope-anchored review. pohe-agent analyze <repo> --pattern E08 anchors the cue to its enclosing declaration and checks a separate operational signature for all four pointcut dimensions. Missing roles and negative-control risks remain visible; a structural match is not E2/E3. Calibration report ↗

Repository files

Up to five public files retrieved by the frozen cue and pinned to the audited commit. A match is a place to inspect, not a confirmed implementation.

No repository file was retrieved for this record.

The frozen panel returned no cue realization, so the record remains specification-only. No repository link is invented to fill the gap.

Evidence & provenance

Each layer answers a different question. Do not substitute retrieval for review.

Specification only

No official executable implementation was located in the supplied study. The paper anchor motivates the record; it is not an implementation claim.

Review boundary: this candidate and its specification come from an AI-assisted study. Independent human validation remains open.

ANY-EVIDENCE FAMILY COVERAGE0/36

0.0% of repository families in the frozen panel have at least one current cue, historical cue, or canonical source layer. This is not confirmed prevalence.

Current lexical-cue repositories: 0 / 41

Paper anchorAdaRSI paper (arXiv:2610.20474)

AdaRSI paper describes choosing harness adaptation versus policy training.

Exploratory retrieval cue Regex / v0.1.0

(?i)\b(?:adaptation[_-]?target|weight[_-]?adaptation|on[_-]?demand[_-]?adaptation)\b

This expression is a discovery instrument only. A match is a retrieval cue—not a confirmed implementation, pointcut match, or contract proof.