Sov Research — Knowledge Ring The Researcher

A Character Sheet for Knowledge Synthesis
Arc · 22 March 2026

I. The Problem of Research

Systems generate vast amounts of code but produce almost no self-knowledge. This is the paradox of modern software: the artifact grows, the documentation atrophies. Every commit adds structure. Every sprint defers explanation. The system becomes more capable and less understood in the same motion, until the gap between what it does and what anyone can say about what it does becomes architecturally load-bearing — a void that shapes every decision made in its proximity.

Documentation is an afterthought — written by humans who don’t have time, about systems they didn’t fully explore, for audiences they can’t predict. The best documentation captures what the author knew at the moment of writing. The worst captures what the author wished were true. Neither captures what the system is: the full topology of its types, the actual flow of its data, the invariants it enforces, the constraints it violates, the mathematical signatures buried in its pure functions. To know a system is to survey it completely, and humans do not survey completely. They sample. They extrapolate. They get tired.

The Researcher proposes an alternative: automated knowledge synthesis. Not documentation generation — that is a solved problem and an insufficient one. Synthesis. A character that surveys, analyzes, cross-references, validates, and publishes — producing Silicon Cartography papers with the same rigor a human researcher would, but with perfect access to the source. The Researcher does not guess what a function does. It reads the function. It does not assume a test passes. It reads the test result. It does not infer the topology. It queries the topology engine. Every claim in a Silicon Cartography paper is grounded in evidence extracted from the system itself.

The system that cannot describe itself is the system that cannot be trusted. The Researcher exists to close the gap between what the system does and what can be said about what the system does.

The ambition is not modest. The Researcher aims to produce papers of publication quality — complete with mathematical signatures, test evidence, structural diagrams, and provenance chains — about any node, ring, or artifact in the sovOS mesh. The Eco Stone paper was the proof of concept. This paper is the self-portrait: the Researcher documenting its own architecture, its own constraints, its own gaps. If it can describe itself honestly, it can describe anything.

II. The Research Workflow

The Researcher operates through a six-phase pipeline. Each phase maps to a scroll, a set of stages, and a spell on the character sheet. The pipeline is sequential — each phase feeds the next — but individual phases can be invoked independently when partial research is sufficient.

Phase 1: Survey

The Researcher identifies its target and gathers structural context. It calls sov_describe to understand the target’s identity, sov_topology to map its position in the mesh, and sov_config_read to extract its declared configuration.

Phase 2: Analyze

The Researcher reads source code with prism to locate function definitions, type declarations, and stage implementations. It then reads the Rust source directly and extracts signatures and relationships.

Phase 3: Cross-Reference

Findings are mapped to the 16 CC sections and mathematical signature candidates are flagged (cosine, Euclidean distance, exponential decay, running mean, cryptographic hashing).

Phase 4: Synthesize

The compose_section stage uses SpellSource::Llm to transform structured findings into narrative while preserving provenance tags.

Phase 5: Validate

Equations are checked against implementations, test counts against cargo test evidence, and structural claims against topology output.

Phase 6: Publish

The paper is rendered as self-contained HTML, index is updated, and governance submission occurs through SAGE.

The Pipeline

III. The Character Sheet

The Researcher is a Legendary-tier character in the sovOS Character Creator system. It inhabits the Knowledge Ring. Its purpose is singular: to produce Silicon Cartography papers that document the system with formal rigor.

The character sheet is not decorative. Every entry corresponds to an enforced capability or constraint in the runtime.

A character sheet is a contract. The system enforces what the sheet declares.

IV. The Spell System

Spells are the Researcher’s executable capabilities. Each spell maps to a scroll, and each scroll orchestrates a sequence of stages.

enum SpellSource {
    Native,   // Compiled Rust stage - deterministic, auditable
    Scroll,   // Interpreted pipeline - scroll orchestrates stages
    Granted,  // From an Eco Stone - capability inherited from artifact
    Llm,      // LLM-mediated - requires external inference (synthesis only)
}

V. The 14 Stages

Every stage is a pure function. Given the same input and same state, it produces the same output.

The pipeline is a directed acyclic graph with no cycles and no back-edges. Research flows forward.

VI. The Seven Scrolls

VII. The Eco Stone

The Researcher carries an Eco Stone. It is an agent-bound stone — StoneBinding::Agent { agent_id: "researcher" } — that accumulates research memory across every paper the Researcher produces.

After each paper is published, findings are absorbed into the stone’s manifold. Every type discovered, equation identified, and relationship traced becomes a 768D memory. The manifold densifies around frequently researched domains.

This compound effect means later papers can retrieve prior findings before full survey completes, enabling cross-domain insight as a geometric side effect.

The stone densifies around what the Researcher studies. After enough papers, the stone knows the system well enough to illuminate connections the Researcher has not yet looked for.

VIII. Items and Inventory

The Researcher carries five items. Each is a CC Object with defined effects that grant capabilities.

Object {
    shape_ref: "EcoStone",
    tier: Mythic,
    binding: Agent { agent_id: "researcher" },
    effects: [
        GrantMemoryAccess("canon/researcher_memory", "rw"),
        GrantStageAccess("eco.*"),
        GrantSpell("reflect"), GrantSpell("absorb"),
        GrantSpell("project"), GrantSpell("attune"),
        GrantSpell("fuse"),
    ],
}

Object {
    shape_ref: "Codex",
    tier: Rare,
    effects: [
        GrantKnowledge("silicon_cartography_template"),
        GrantKnowledge("cc_section_schema"),
        GrantKnowledge("equation_format"),
    ],
}

Object {
    shape_ref: "Ledger",
    tier: Uncommon,
    effects: [
        GrantMemoryAccess("canon/*", "r"),
        GrantMemoryAccess("shadow/researcher", "rw"),
    ],
}

Object {
    shape_ref: "Lens",
    tier: Rare,
    effects: [
        GrantSignalVisibility("hermes/*"),
        GrantSignalVisibility("knowledge/*"),
        GrantSignalVisibility("eco/*"),
    ],
}

Object {
    shape_ref: "Seal",
    tier: Uncommon,
    effects: [
        GrantAuditWrite("research/*"),
        GrantGovernanceAccess("sage/submit"),
    ],
}

IX. Constraints and Governance

The Researcher is deliberately limited to preserve output integrity.

Constraint 1: Read-Only System

Cannot modify source/config/stage state outside its scoped research paths.

Constraint 2: Cannot Build

Cannot invoke build stages; research and engineering are separated by design.

Constraint 3: SOV_ROOT Boundary

Cannot access files outside workspace boundary.

Constraint 4: Journal All Actions

All actions are journaled to Owl with timestamped traces.

Governance: SAGE Gate

Publishing requires governance approval for schema, evidence, and provenance completeness.

X. What the Researcher Cannot Do

Cannot Write Code

The Researcher reports implementation gaps but cannot fill them.

Cannot Run Tests

It reads test evidence but cannot execute cargo test directly.

Synthesis Requires LLM Mediation

compose_section is non-deterministic by design; facts remain sourced.

Cannot Validate Runtime Behavior

Papers capture static code and test evidence, not live production behavior.

Cannot Research Itself Without External Context

The self paper is bootstrap specification before full runtime availability.

XI. The Bootstrap Problem

This paper was written through human-agent collaboration because the full Researcher runtime is not yet live. The stages are specified, the scrolls declared, and the stone forged, but execution is still in bootstrap phase.

Every self-referential system faces this pattern: first iteration manual, second iteration uses the first, nth iteration becomes effectively magical through accumulation.

The bootstrap problem is temporary. The first paper is a gift. Every paper after is geometry.

The first paper is a gift. Every paper after is geometry.