---
title: "native-octopus-deputy — Mova code model"
description: "Mova code model of models/native-octopus-deputy.groupant: components with their labels, obernants and the model text."
url: "https://groupants.elamur.ai/free/en/mova/models/native-octopus-deputy.groupant/"
---
Elamur: [Mova](https://md.mova.elamur.ai/free/en/) · [Dictionary](https://md.dictionary.elamur.ai/free/en/) · [Research](https://md.research.elamur.ai/free/en/) · **Groupants**

This page as HTML: <https://groupants.elamur.ai/free/en/mova/models/native-octopus-deputy.groupant/>

# native-octopus-deputy

The groupant beside models/native-octopus-deputy.groupant in the Mova repository: what the code computes, what it never does, how to go back from its outputs, its invariants, cost and checks.

## Facts

- **kind:** groupant_group
- **status:** partial source bound model
- **bound sources:** 9
- **repository folder:** [`models/native-octopus-deputy.groupant`](https://github.com/ElamurAI/mova-free/tree/main/models/native-octopus-deputy.groupant)
- **repository commit:** `855da7bd4a7b`

> **Implementation material.** A Mova code model describes this code at one source revision: partial models, finite checks and their stated scope. It is not a reviewed general exposition, and its labels are shown as written, never raised.

## Components

| component | label | where |
|---|---|---|
| **approximant** | partial source derived | [Forward](#forward) |
| **ast** | not materialized |  |
| **composition** | partial source derived | [Composition](#composition) |
| **cost** | partial source derived | [Cost](#cost) |
| **descriptant** | partial source derived | [Descriptant](#descriptant) |
| **invariant** | partial source derived | [Invariant](#invariant) |
| **tests** | finite evidence | [Evidence](#evidence) |

## Obernants

The typed relations around the forward model: the behaviour complement (what the code never does), the reverse reconstruction (every input compatible with an output) and specialised views.

| id | type | label | where |
|---|---|---|---|
| `behavior_complement` | behavior_complement | partial source derived finite checks | [behavior_complement](#behavior-complement) |
| `reverse_reconstruction` | reverse_reconstruction | partial source derived finite checks | [reverse_reconstruction](#reverse-reconstruction) |
| `handoff_admission_exclusion` | handoff_admission_exclusion | partial source derived finite checks | [handoff_admission_exclusion](#handoff-admission-exclusion) |

## Model text


### Descriptant
The abstract deputy interprets ordinary English through actual Mova syntax observations and configured capability rules. Concrete actors and assignments are external configuration. Two supplied adapters establish a bounded read-only starting point; additional installed handlers do not imply general understanding.

### Forward
P -> native UD/frame observations -> all-sentence admission -> named capability plan -> durable reservation -> installed handler calls -> source/output-bound receipt. A refusal instead produces a review-only highest-priority diagnostic candidate with the actual syntax and explicit pragmatic-engine gap.

### Invariant
No planning refusal reaches a handler. No diagnostic candidate is automatic gold or training authorization. Hash identity preserves bytes and provenance, not semantic truth. Completed invocation reuse preserves an old observation rather than asserting freshness.

### Cost
CLI/model loading plus bounded native parsing/configuration matching and at most eight adapter calls. Concrete adapter latency and end-to-end quality remain separate; no speedup theorem.

### Composition
Correct parsing alone does not prove correct intent binding. Correct typed plans alone do not prove handler semantics. Correct transport alone does not establish health. Configured aliases/operations and returned values remain auditable in trace. Future extensions require their own domain and checks.

### behavior_complement
Domain: the supplied CLI/library/adapters under cooperative configuration. Excluded transitions include partial execution of a refused multi-sentence prompt and promotion of raw failure to a learned parser sample. Counterexample: a valid Crab request plus unsupported mutation must remain wholly refused. Evidence: finite negative controls and source wiring.

### reverse_reconstruction
The same concrete capability plan can arise from multiple natural paraphrases. Native traces retain actual prompt identity, but neither plan nor status response identifies a unique intended meaning or system history. Counterexample: polite and imperative requests yield equal plans. Evidence: finite native controls.

### handoff_admission_exclusion
Domain: generated review candidates. Highest priority does not satisfy privacy, held-out or semantic admission. Candidate hashes and actual native observations cannot fill missing pragmatic inference. Counterexample: unresolved “its logs” has valid UD but no referent; it is review material, not a corrected annotation. Evidence: actual source-bound development probe; reviewed downstream adaptation remains outside this groupant.

### Evidence
2026-11-02: four focused Rust tests passed with ten natural positive and eight refusal prompts, mocked Unix idempotence/uncertainty controls and lossless native-feature serialization. Real configured status and combined log/status requests completed. The trusted current stderr was empty: bytes_read=0, healthy=null. No training, held-out test, general capability or health claim.

### Epoch 2026-11-04 (Sea Spider live: singleton and callers)

New child octopus-animal-relay: the built-in handler animal_ask relays quoted English to a configured animal socket as a declared caller; the socket envelope carries an optional from.