DISCOVERY SUBSTRATE 4 CLAUDE CODE (DS4CC)

DISCOVERY SUBSTRATE 4 CLAUDE CODE — a strict Pareto dominance interface for research-style agent coordination.

Reference implementation of the Umwelt-Frontier Orchestrator (UFO)

exclusion-topology — the admissibility boundary that filters dominated moves while routing novel proposals past local attractors · Umwelt-Frontier Orchestrator (UFO) reference implementation · Free & open source · Windows reference daemon

Dominance gate

Figure: Pareto membrane — each proposal must be non-regressive on every axis and strictly improve at least one. Dominated moves are excluded at the boundary.

Research substrate interface

Velocity

Heterogeneous proposal fan-out across symbolic, UI, and shell action channels — including directional navigation that resolves to arrow keys. Table 1 row 1: the GTA jetpack motif marks proposals that escape local optima by crossing novelty-rich transition regions while preserving admissibility.

Umwelt / Axis Governance

R2 codifies axes declared externally: environment state, risk surface, and objective framing. A revolving boundary tracks permitted directions, functioning as the exclusion membrane that filters context drift while preserving the agent’s operational topology.

No-goal Search

Inputs become proposals, not commands. The run explores admissible variants without collapsing them into a single terminal objective.

Pareto Filter

R1 enforces a strict dominance gate. Cursor motion with analog wobble serves as perturbation search; when movement lands in the admissible set, visual ripple marks acceptance—an operational test for non-dominated progression.

Mutation Testing

Mutator probes inject adversarial perturbations across state channels, analogous to matrix-rain stress sequences. At 76.9% kill rate, these sweeps validate robustness by forcing transitions that challenge brittle trajectories before deployment.

Saturation Exit

R4 triggers saturation-based exit when no admissible move crosses the exclusion boundary. Rumble-to-whip-to-border transitions encode the signal: search front saturated, axis redefinition or higher-order perturbation required.

Pattern specification

Eq 1 — Pareto admissibility:
∀ a_i ∈ A : s_i(m) ≥ s_i(x_t) ∧ ∃ a_j ∈ A : s_j(m) > s_j(x_t)

R1: Strict dominance gate — a move m is admitted only if it is weakly better on all axes and strictly better on at least one; dominated proposals are filtered at the exclusion boundary.
R2: Axis-first governance — axes are declared externally and remain stable across a run; no implicit scalar collapse is permitted mid-search.
R3: No terminal scalar utility — the orchestrator carries no single objective to minimise or maximise; search continues until saturation, not until a target value is reached.
R4: Saturation-based exit — termination is triggered when no admissible move crosses the exclusion boundary; axis redefinition or higher-order perturbation is required to resume.

Scope & claims

✓ Substrate scope: wide-parallel synthesis tasks where heterogeneous proposals are fan-outed across symbolic, UI, and shell channels under axis-first governance and saturation-based exit.
✓ Internal evidence: mutation kill-rate 76.9% (10/13 mutants killed) — adversarial sweeps validate non-regressive trajectory robustness before deployment.
not claimed End-to-end superiority over all architectures (paper §1.3).
not claimed Universal replacement for graph-native or single-agent reasoning workflows (paper §1.3).

9-node lineagepaper §4 / Table 2
- The Ratchet
  - no-goal search
  - saturation exit
- Pareto Dominance
  - antimetabole
  - exclusion-topology
- Quality Recognition / Pirsig
  - Umwelt / von Uexküll
  - novelty search