Skip to content
Hearth & Code / independent studio research, making, and public record
Hearth & Code

Research, making, and a long way back into the work.

Hearth & Code / local candidate

Exploring a candidate one-person research-lab and creative-hub posture for tools, evidence, and connected work.

Hearth & Code is testing this candidate posture through small proofs, explicit limits, and a public record that stays separate from the private working practice.

The posture does not imply a staff, accredited institution, dedicated facility, commissioned program, mature platform, or validated outcome.

Orientation signal

Current inquiry
How can source-aware tools make complex work easier to inspect, return to, and correct?
Released public records
0 records are released for direct public action.
Candidate inventory
No candidate count is presented as a current public signal.
Next gate
Review the evidence, limits, relations, and page composition before any public release.

Practice / responsibility

One person is responsible for the direction and the release.

Candidate research-lab posture: an operating commitment.

It means framing focused questions, building or testing small proofs, recording evidence and uncertainty, and returning results for review.

Candidate creative-hub posture: a connective center.

It joins selected software, publishing, visual, musical, and narrative work without claiming that every medium is active, available, or mastered.

Person-led does not erase assistance.

Human intention, selection, judgment, verification, and release remain identifiable. Material tools and collaborators are disclosed for each artifact.

The scale stays literal.

This candidate posture is not a team, collective, membership community, staffed studio, facility, accredited lab, or source of generalizable results.

Current investigations

Questions and proofs, with the edges still visible.

Each item below is a qualified summary of a candidate record in the single rendered Library. Canonical authority remains in the bound Hub sources; the complete evidence label, limit, non-claim, and contribution boundary stay with that record.

  • Experimental · Software prototype

    Hearth & Code Workbench pre-alpha

    Question
    What can a bounded local-first workroom make inspectable?
    Small proof or method
    A pre-alpha profile-evaluation slice with deterministic mock and receipt paths.
    Evidence
    Exact public-repository observation
    Limitation
    No live model, agent runtime, production CoreStore, semantic retrieval, cloud sync, or current CI result is established here.
    Does not establish
    Not a finished platform, launched product, autonomous-agent system, or demonstrated cognitive benefit.
    Observed
    2026-08-10
    Action state
    Source link held for review — Exact destination release is held.
    Candidate correction route
    /method/#corrections

  • Experimental · Software and workflow proof

    Context Forge workflow proof

    Question
    How can agent-assisted procedures remain source-governed and reviewable?
    Small proof or method
    Workflow schemas, deterministic simulation, synthetic fixtures, and explicit review gates.
    Evidence
    Exact public-repository observation
    Limitation
    No autonomous execution, provider invocation, credential handling, deployed service, or canonical write is established.
    Does not establish
    Not a production agent platform or autonomous orchestration service.
    Observed
    2026-08-10
    Action state
    Source link held for review — Exact destination release is held.
    Candidate correction route
    /method/#corrections

  • Experimental · Knowledge-tool proof

    Bounded retrieval and ContextPack proof

    Question
    Can a ContextPack remain small, cited, and drift-aware?
    Small proof or method
    A lexical-hash retrieval proof with source allowlists and a dated five-test receipt.
    Evidence
    Exact public-repository observation
    Limitation
    The proof uses lexical-hash retrieval only; its receipt is dated and remains unverified.
    Does not establish
    Not semantic intelligence, general memory, autonomous injection, or production readiness.
    Observed
    2026-08-10
    Action state
    Source link held for review — Exact destination release is held.
    Candidate correction route
    /method/#corrections

  • Proposed · Taxonomy and design artifacts

    Knowledge-governance pilot artifacts

    Question
    Which vocabulary and provenance constraints make knowledge work inspectable?
    Small proof or method
    Public pilot taxonomy and design artifacts; no installed service is claimed.
    Evidence
    Exact public-document observation
    Limitation
    A proposed approach and pilot artifacts, not an installed service or complete ontology.
    Does not establish
    Not proof that all Hearth & Code records are governed or validated.
    Observed
    2026-08-10
    Action state
    Source link held for review — Exact destination release is held.
    Candidate correction route
    /method/#corrections

  • Experimental · Authored N-of-one field note

    Agent-assisted procedure-design field note

    Question
    What does one evolving agent-assisted procedure look like in practice?
    Small proof or method
    A dated, self-reported N-of-one field note with assistance disclosed.
    Evidence
    Exact public-authored observation
    Limitation
    Self-reported; no benchmark, external validation, productivity result, or reproducibility proof.
    Does not establish
    Not validated agentic-engineering expertise or achieved mastery.
    Observed
    2026-08-10
    Action state
    Source link held for review — Exact wording and authorship review are required.
    Candidate correction route
    /method/#corrections

Tools / prototypes

Inspectable candidates at their recorded maturity.

Software prototype · Experimental

Hearth & Code Workbench pre-alpha

One inspectable local-first pre-alpha profile-evaluation workroom implemented with Tauri, TypeScript, and Rust, including a deterministic mock path and bounded receipt and persistence proof.

Evidence: Exact public-repository observation

Limitation: No live model, agent runtime, production CoreStore, semantic retrieval, cloud sync, or current CI result is established here.

Does not establish: Not a finished platform, launched product, autonomous-agent system, or demonstrated cognitive benefit.

Contribution: Public repository evidence discloses AI-assisted development under human direction and review.

Observed: 2026-08-10

Status: Source link held for review — Exact destination release is held.

Candidate correction route: /method/#corrections

Software and workflow proof · Experimental

Context Forge workflow proof

A pre-release proof for source-governed workflow schemas, graph validation, deterministic simulation, review gates, synthetic fixtures, and a local interface.

Evidence: Exact public-repository observation

Limitation: No autonomous execution, provider invocation, credential handling, deployed service, or canonical write is established.

Does not establish: Not a production agent platform or autonomous orchestration service.

Contribution: The public package identifies Scott as author and Hearth & Code as repository owner; this does not imply a team.

Observed: 2026-08-10

Status: Source link held for review — Exact destination release is held.

Candidate correction route: /method/#corrections

Knowledge-tool proof · Experimental

Bounded retrieval and ContextPack proof

A local deterministic bounded-retrieval and ContextPack proof using source allowlists, citations, hashes, drift detection, and a dated five-test receipt.

Evidence: Exact public-repository observation

Limitation: The proof uses lexical-hash retrieval only; its receipt is dated and remains unverified.

Does not establish: Not semantic intelligence, general memory, autonomous injection, or production readiness.

Contribution: Public-safe derivative of independently directed engineering work; private source material remains excluded.

Observed: 2026-08-10

Status: Source link held for review — Exact destination release is held.

Candidate correction route: /method/#corrections

Local static-site source · Current state unknown

Landing source implementation

A local static Astro implementation exists for the landing and principal orientation routes at the bound commit.

Evidence: Bound local-source observation

Limitation: Remote currentness, deployment, public reachability, and current behavior remain unchecked.

Does not establish: Not proof of publication, readership, accessibility conformance, or public availability.

Contribution: Checked-in source is observable; precise human and AI contribution is not established by this evidence item.

Observed: 2026-08-10

Status: Information only — Remote and deployment currentness are unchecked.

Candidate correction route: /method/#corrections

Knowledge library

One rendered collection, derived from bound Hub sources.

The Library is the page's single rendered record source. Canonical authority remains in the bound Hub sources, and no candidate action below is an enabled public link.

Practice orientation

Organizational orientation · Current at the observed revision

Hearth & Code practice orientation

Candidate orientation for a one-person, person-led research and creative-technology practice moving toward a separately gated research-lab and creative-hub posture.

Evidence
Canonical organizational-source candidate
Limitation
The lab and creative-hub wording is not yet a canonical or released public identity.
Does not establish
Not a team, accredited institution, facility, commissioned program, community, mature product company, or achieved-expertise claim.
Contribution
Scott remains the responsible human author and reviewer; collaborators and material tools remain specific to each artifact.
Observed or updated
2026-08-10
Action state
Information only — Candidate posture only.
Candidate correction route
/method/#corrections
Return to Knowledge Library

Software and workflow proofs

Software prototype · Experimental

Hearth & Code Workbench pre-alpha

One inspectable local-first pre-alpha profile-evaluation workroom implemented with Tauri, TypeScript, and Rust, including a deterministic mock path and bounded receipt and persistence proof.

Evidence
Exact public-repository observation
Limitation
No live model, agent runtime, production CoreStore, semantic retrieval, cloud sync, or current CI result is established here.
Does not establish
Not a finished platform, launched product, autonomous-agent system, or demonstrated cognitive benefit.
Contribution
Public repository evidence discloses AI-assisted development under human direction and review.
Observed or updated
2026-08-10
Action state
Source link held for review — Exact destination release is held.
Candidate correction route
/method/#corrections
Return to Knowledge Library

Software and workflow proof · Experimental

Context Forge workflow proof

A pre-release proof for source-governed workflow schemas, graph validation, deterministic simulation, review gates, synthetic fixtures, and a local interface.

Evidence
Exact public-repository observation
Limitation
No autonomous execution, provider invocation, credential handling, deployed service, or canonical write is established.
Does not establish
Not a production agent platform or autonomous orchestration service.
Contribution
The public package identifies Scott as author and Hearth & Code as repository owner; this does not imply a team.
Observed or updated
2026-08-10
Action state
Source link held for review — Exact destination release is held.
Candidate correction route
/method/#corrections
Return to Knowledge Library

Knowledge-tool proof · Experimental

Bounded retrieval and ContextPack proof

A local deterministic bounded-retrieval and ContextPack proof using source allowlists, citations, hashes, drift detection, and a dated five-test receipt.

Evidence
Exact public-repository observation
Limitation
The proof uses lexical-hash retrieval only; its receipt is dated and remains unverified.
Does not establish
Not semantic intelligence, general memory, autonomous injection, or production readiness.
Contribution
Public-safe derivative of independently directed engineering work; private source material remains excluded.
Observed or updated
2026-08-10
Action state
Source link held for review — Exact destination release is held.
Candidate correction route
/method/#corrections
Return to Knowledge Library

Knowledge and method artifacts

Taxonomy and design artifacts · Proposed

Knowledge-governance pilot artifacts

Public pilot taxonomy and design artifacts for controlled vocabulary, integrity rules, provenance, and source-aware knowledge governance.

Evidence
Exact public-document observation
Limitation
A proposed approach and pilot artifacts, not an installed service or complete ontology.
Does not establish
Not proof that all Hearth & Code records are governed or validated.
Contribution
Reduced public derivatives with private identities and runtime details excluded.
Observed or updated
2026-08-10
Action state
Source link held for review — Exact destination release is held.
Candidate correction route
/method/#corrections
Return to Knowledge Library

Authored N-of-one field note · Experimental

Agent-assisted procedure-design field note

A dated N-of-one field note documents an evolving practice of goal contracts, bounded delegation, disjoint ownership, review gates, and receipts with AI assistance disclosed.

Evidence
Exact public-authored observation
Limitation
Self-reported; no benchmark, external validation, productivity result, or reproducibility proof.
Does not establish
Not validated agentic-engineering expertise or achieved mastery.
Contribution
Scott owns the practice description, judgment, and publication decision; exact first-person reuse requires separate review.
Observed or updated
2026-08-10
Action state
Source link held for review — Exact wording and authorship review are required.
Candidate correction route
/method/#corrections
Return to Knowledge Library

Publishing and web source

Journal implementation and corpus · Current at the observed revision

Field Journal source corpus

An Astro and React source corpus with six source entries, structured frontmatter, RSS, reading controls, and AI-provenance fields at the checked revision.

Evidence
Exact public-repository observation
Limitation
Deployed blog availability was not directly checked for this evidence set.
Does not establish
Journal publication does not independently validate article claims or establish research expertise.
Contribution
Each article retains its own author, assistance disclosure, and claim map.
Observed or updated
2026-08-10
Action state
Source link held for review — Deployed route remains unchecked.
Candidate correction route
/method/#corrections
Return to Knowledge Library

Local static-site source · Current state unknown

Landing source implementation

A local static Astro implementation exists for the landing and principal orientation routes at the bound commit.

Evidence
Bound local-source observation
Limitation
Remote currentness, deployment, public reachability, and current behavior remain unchecked.
Does not establish
Not proof of publication, readership, accessibility conformance, or public availability.
Contribution
Checked-in source is observable; precise human and AI contribution is not established by this evidence item.
Observed or updated
2026-08-10
Action state
Information only — Remote and deployment currentness are unchecked.
Candidate correction route
/method/#corrections
Return to Knowledge Library

Knowledge atlas

Literal relationships among the same Library records.

The Atlas is a navigation projection, not another source. Its circuitry may disappear without removing any relationship or route.

Methods / evidence / maturity

A return loop for claims that need to stay revisable.

  1. Ask. Name a focused question and the boundary of the inquiry.
  2. Build or test. Use the smallest proof that can expose useful evidence.
  3. Record. Keep the evidence, assistance, uncertainty, limitation, and date visible.
  4. Return. Review the result, correct the claim, and preserve what changed.

Maturity belongs to each record, not to the whole practice.

Evidence and limitation travel together.

Each record states an evidence label, a limitation, what it does not establish, its observation or update date, and a literal candidate correction route: /method/#corrections.

Human and tool contribution stays inspectable.

Direction, selection, judgment, verification, and release remain human responsibilities. Material AI assistance, collaborators, and tool contributions are disclosed at the artifact level.

This operating commitment does not establish scientific validity, universal compliance, reproducibility, efficacy, accreditation, or accessibility conformance.

Field Journal / builder / return

Follow the record, inspect responsibility, or return to a question.

Field Journal

Candidate source records and field notes remain visible at their observed maturity; deployed availability and exact article wording remain separately gated.

Builder responsibility

Hearth & Code is one-person and person-led. Scott retains responsibility for direction, judgment, review, and release; assistance and collaboration stay artifact-specific.

Return

Candidate correction route: /method/#corrections