---
title: Knowledge Production System
section: Ora — System papers
status: review
description: The part of Ora that produces the substrate the rest of the system runs on — how raw inputs become provenance-weighted, retrievable knowledge.
authors:
  - The Ora Foundation
downloads:
  md: /papers/white/knowledge-production-system.md
license: https://creativecommons.org/publicdomain/zero/1.0/
---

# Knowledge Production System

## Overview

The Knowledge Production System is the part of Ora that produces the substrate the rest of the system runs on. The other systems supervise (Strategic Supervision), enforce (Coordination Layer), shape (Matrix Lifecycle), and organize information flow (Information Lifecycle). This system produces the *content* — the user's MindSpec, which is the values substrate Ora reads to make value-aware decisions; the atomic notes that fill the engram vault and provide the knowledge substrate Ora's RAG and IIF cross-pool math read; and the conceptual frame that explains what creativity is in this system and how the apparatus produces it.

Three components compose the system. The **MindSpec Interview Framework** produces complete MindSpec specifications through tiered interactive assessment — a 66-entry library across 11 families, a three-stage assessment instrument (66 portraits, 84 dyad scenarios, 35 multi-commitment scenarios, 17 constitutional pressure-tests), an inference layer that handles incompatibility adjustments and selfishness-coefficient weighting, and a learning architecture with six drift-prevention safeguards. The framework runs in three tiers — Tier 1 ephemeral (a single session's MindSpec for a specific assistance context); Tier 2 persistent task (a MindSpec scoped to a particular kind of ongoing work); Tier 3 personal thinking partner (the user's full self-MindSpec). It produces the canonical mind.md file (4,000–6,000 words for Tier 3, all nine sections, weight-6+ commitments at 100–200 words each), an empty ledger.md, and an empty modifications.md.

The **Knowledge Artifact Coach (KAC v6.0)** produces atomic notes — the smallest knowledge units in Ora's engram corpus. Each atomic carries a single discrete claim, the evidence supporting it, the relationships connecting it to other atomics (via the 13-type relationship taxonomy — `supports`, `contradicts`, `extends`, `qualifies`, `instantiates`, `refines`, `presupposes`, `competes-with`, etc.), and the metadata that supports retrieval (provenance markers per the YAML schema). The framework runs in coaching mode — interactive elicitation that helps the user produce well-formed atomics — and in batch mode for processing source documents into engrams via the Document Processing pipeline. KAC is the framework that lets the user (or the AI on the user's behalf) produce knowledge that the rest of the system can index, retrieve, and combine.

The **Reference — Creativity from Knowledge and Values** is the conceptual frame. It articulates Ora's understanding of what creativity is in this system: a composite act with three steps — generative search through combinatorial space (combining things from the knowledge substrate), value-filtered selection (the values substrate scores combinations for fit), and action selection (the recognition that fires when a high-value combination is registered as creative). The Reference frames generative search and value-filtering as automatable; action selection is where "what remains human" lives. The Inception and Incubation Framework (covered in [Matrix Lifecycle System](/papers/matrix-lifecycle-system)) refined this in 2026-05-08 by collapsing the composite to a single irreducible step — recognition. Generation is automatable; value-filtering is automatable; recognition is not. The Reference and the IIF refinement together provide the conceptual story that makes the Knowledge Production System's purpose legible — produce the substrate; let the system do the combinatorial work; preserve the recognition residue as the empirical signature of irreducibly human contribution.

What the system does that no individual component does is produce the **substrate the entire rest of Ora reads**. MindSpec alone is a personal values document. Atomic notes alone are scattered claims. The Creativity reference alone is an essay. Together, they produce the substrate that PEF reads when supervising matrices (the Excluded Outcomes draw on the user's values), that MOM reads when classifying ideas (the four-test classification respects the user's commitments), that IIF reads when running Mode 1 generation (the value-alignment math is MindSpec × vault matrix index), that the Coordination Layer enforces (the locks reflect the user's values), that the Information Lifecycle System organizes (the corpora and outputs carry the user's substantive content). The substrate is the precondition for everything else — without MindSpec, value-alignment cannot run; without atomic notes, the knowledge substrate is empty; without the Creativity reference, the system has no story for what it is doing when it produces candidates the user recognizes.

## Systemic context

The Knowledge Production System is the most upstream system in Ora — its outputs feed every other system. The [Strategic Supervision System](/papers/strategic-supervision-system) reads MindSpec for value alignment in the Resolution Statement Objectivity Protocol's Near-Miss Elicitation and in the Excluded Outcomes derivation. The [Coordination Layer](/papers/coordination-layer) supervises but does not produce the system's artifacts — Process Coherence supervises corpus writes (KAC-style atomics may flow into corpora; CFF-formalized corpora may consume from KAC engrams). The [Matrix Lifecycle System](/papers/matrix-lifecycle-system)'s Inception and Incubation Framework reads MindSpec values directly in Mode 1's value-alignment math; the Spark Recognition Corpus accumulates the empirical evidence that Recognition is the irreducible step (the central claim of the Creativity reference as refined by IIF). The [Information Lifecycle System](/papers/information-lifecycle-system) composes with the system's outputs — KAC engrams flow into CFF-formalized corpora; OFF-rendered outputs may render from KAC engram subsets; Document Processing produces engrams via KAC and resource chunks from source documents per the YAML schema's provenance hierarchy.

## Constituent components

**MindSpec Interview** — the values-substrate producer. Single-file operational specification (single-file architecture restored 2026-05-09 to remove distribution dependencies). Sections: §I Governing Principles; §II The Library (66 entries across 11 families with full schema fields, near-enemies, direct-oppositions, scale anchors); §III Schema Specification; §IV Three-Stage Assessment Instrument (66 portraits, 84 dyad scenarios, 35 multi-commitment scenarios, 17 constitutional pressure-tests); §V Inference Layer (incompatibility adjustments, selfishness-coefficient weighting, adjustment transparency); §VI Learning Architecture (six drift-prevention safeguards); §VII Tier-Dependent Interview Flows; §VIIA Stage 2A Life-Context Direct Pass; §VIII Evaluation Criteria; §IX Named Failure Modes; §X Execution Commands; §XI Registry Entry. Universal instrument applies to self, persistent agents, and fictional characters — the MindSpec produced for the user, for an AI agent the user wants to instantiate, or for a fictional character the user wants to model are all produced by the same instrument with the same sections. v0.2.2 universality-audit corrections applied (26 default values recalibrated to general-population median).

**Knowledge Artifact Coach (KAC v6.0)** — the atomic-note producer. Coaches the user (or operates batch-style on source material) to produce well-formed atomic notes. Each atomic carries a single discrete claim (the load-bearing rule — "one claim per atomic"), the evidence supporting the claim, the 13-type relationship links to other atomics (`supports`, `contradicts`, `extends`, `qualifies`, `instantiates`, `refines`, `presupposes`, `competes-with`, `bridges`, `compresses`, `decomposes`, `disambiguates`, `cites`), and the YAML frontmatter that supports retrieval (provenance markers per the schema — since rev 5.2 a kept engram weighs the same regardless of authorship; the `ai-derived` marker is recorded for the cleaning framework, not for weighting). The framework runs in interactive coaching mode for the user-authored case and in batch mode for the source-document case (where Document Processing chunks the source and KAC distills atomics from the chunks). The 13-type relationship taxonomy is the load-bearing structural decision — atomics with explicit typed links produce a knowledge graph that the meta-layer's semantic-similarity engine can traverse semantically rather than purely lexically.

**[Creativity from Knowledge and Values](/papers/creativity-from-knowledge-and-values)** — the conceptual frame. Articulates the Knowledge Production System's understanding of creativity as a composite act with three steps — generative search through combinatorial space, value-filtered selection, action selection. Frames generative search and value-filtering as automatable; locates "what remains human" in action selection. The Reference's "Refinement: Creativity Collapses to Recognition" section (added 2026-05-08 alongside the IIF landing) sharpens the claim — recognition is not just a step in the composite; it is the residue that remains after generation and value-filtering have done all they can do. The "What Remains Human" section sharpens further — the recognition act is the irreducibly human contribution by design, and the IIF Recognition Lock enforces that recognition cannot be silently substituted by automation. The Reference is read by the Inception and Incubation Framework, by MOM (when value-alignment matters in Resolution Statement work), and by users orienting to the system's purpose.

## End-to-end worked example

**Scenario.** A user new to Ora is setting up the Knowledge Production System. They have not yet built a MindSpec; their vault has scattered notes but no atomics; they do not yet have a story for what the system is doing when it produces creative output. Walk through how the system produces the substrate.

**Phase 1 — MindSpec Interview / Tier 3.** The user invokes the MindSpec Interview Framework in Tier 3 mode (full personal thinking partner). The framework runs Stage 1 (66 portrait scenarios — quick reactions to one-sentence portraits of value commitments), Stage 2 (84 dyad scenarios — choices between two value commitments in tension), Stage 2A (life-context direct pass — what the user is currently working on, who matters to them, what's at stake), Stage 3 (35 multi-commitment scenarios — complex situations involving three or more value commitments simultaneously), and the constitutional pressure-tests (17 scenarios that probe the most load-bearing commitments). The framework produces the user's mind.md — 4,800 words, all nine sections (Identity, Constitutional Commitments, Working Commitments, Aesthetic Sensibility, Communication, Memory, Style, Voice, References), with weight-6+ commitments described at 100–200 words each. The mind.md captures: the user is a thinker rather than a builder by orientation; truth-seeking is weight-9; aesthetic craft is weight-7; commercial success is weight-3 with a revisit trigger if it becomes load-bearing; etc. The Inference Layer applies the universality-audit corrections so the defaults reflect the general-population median rather than framework-author residual bias.

The mind.md is written; ledger.md and modifications.md are initialized empty. The user's MindSpec is now the values substrate the rest of Ora can read.

**Phase 2 — KAC interactive coaching for first atomics.** The user has notes from a recent reading of three philosophy-of-attention texts. They invoke KAC in coaching mode and start producing atomics. The framework asks: what's the discrete claim from this passage? What's the evidence? Which existing atomics (none yet — this is the first batch) does this connect to? The user produces the first atomic:

> **Atomic: "Attention is selection at the boundary of awareness."**
> *Evidence:* Citation to William James's *Principles of Psychology* (1890) volume 1, page 403–404; James's framing that attention is the act of taking possession by the mind of one out of several simultaneously possible objects. Page reference and the relevant quote.
> *Relationships:* (none yet — first atomic in the corpus)
> *YAML frontmatter:* `nexus: [philosophy-of-attention]`; `type: engram`; `tags: [atomic]` (distilled from external source material; as a kept engram it weighs 1.0 — authorship no longer modifies weight, per schema rev 5.2).

Over the next two hours, the user produces fourteen atomics. As they accumulate, the relationship typing becomes the load-bearing work — atomic 8 `supports` atomic 3; atomic 11 `contradicts` atomic 6; atomic 14 `bridges` atomics 4 and 9 across two different sources. The 13-type taxonomy is what turns the fourteen atomics from a list into a knowledge graph.

**Phase 3 — Document Processing batch into engrams.** The user has a PDF of a contemporary cognitive-science survey on attention. They invoke Document Processing, which chunks the PDF into resource chunks (P2 retrieval tier) and dispatches the chunks to KAC in batch mode. KAC produces an additional 47 atomics from the chunks, each carrying the chunk's source citation. The relationship-typing pass connects the new atomics to the existing fourteen via the same 13-type taxonomy. The corpus now has 61 atomics across two source authors with explicit cross-source relationships.

**Phase 4 — Reference orientation.** The user reads `Reference — Creativity from Knowledge and Values` and the IIF refinement. They understand the system's story: the atomics they're producing become the knowledge substrate; their MindSpec becomes the values substrate; IIF (when they instantiate it) will run value-alignment math at MindSpec × vault matrix index intersections; the recognitions they make on IIF-surfaced candidates will accumulate into the Spark Recognition Corpus; the residue of recognitions the value model cannot predict will be the empirical evidence for what creativity actually is in this system.

The user can now see what they're doing. The MindSpec captures their values. The atomics build the knowledge substrate. IIF (in the Matrix Lifecycle System) will use both. The Coordination Layer (in the Coordination Layer paper) supervises the workflows that produce more atomics. The Strategic Supervision System (in its own paper) reads MindSpec for value-alignment in supervisory work. The whole system runs on the substrate this system produces.

**Phase 5 — Refinement loop.** Over the next three months, the user continues to add atomics (~200 more). They iterate on MindSpec twice via the Modifications log when life context shifts (a major project ends; a new health context emerges). The Knowledge Production System runs continuously — every reading session adds atomics; every life shift updates MindSpec; the Reference's framing becomes more vivid as the user sees IIF Mode 1 cycles produce sparks that they recognize for reasons MindSpec did not predict (the residue's first entries).

That is the substrate the rest of Ora reads. The system runs in the background of the user's regular work — MindSpec capture is one-time at Tier 3 setup with periodic light updates; KAC is interactive whenever new knowledge is being processed; the Reference is read once at orientation and re-read when the system's purpose needs re-grounding. The substrate accumulates; the rest of Ora reads it; the value of the substrate compounds as it grows.

## How to compose this system

You can run the Knowledge Production System pattern with any AI of your choice. The composition is per-component rather than system-wide because the components produce different artifact types.

**For MindSpec setup**, the prompt is:

> [Paste the Framework — MindSpec Interview specification]
>
> Run the interview at Tier 3 (full personal thinking partner).
>
> [Optional context: anything you want to share up front — recent life context, existing self-understanding, things you're trying to get clearer on.]

The framework runs Stage 1 → Stage 2 → Stage 2A → Stage 3 → constitutional pressure-tests, producing the mind.md at the end. This is heavy — the full Tier 3 interview takes 90–120 minutes. Tier 2 (persistent task scope) is shorter; Tier 1 (single-session ephemeral) is shorter still. For lighter use, start with Tier 1 and let the MindSpec accumulate as the use deepens.

**For KAC atomic production**, the prompt is:

> [Paste the Framework — Knowledge Artifact Coach specification]
>
> I want to produce atomics from this material: [Paste the source — a passage, an article, your own notes from a reading.]
>
> Run KAC in coaching mode. Help me produce well-formed atomics with the 13-type relationship typing.

The framework asks one-claim-per-atomic style questions, helps the user identify the discrete claim, asks for evidence and citation, asks which existing atomics the new one relates to and how (using the 13-type taxonomy). The user produces the atomic; the framework verifies the format; the atomic lands in the engram vault.

For batch mode (a long source document needs to be processed), the prompt is:

> [Paste the Framework — Knowledge Artifact Coach specification]
>
> Run KAC in batch mode against this source: [Paste the document or describe the file.]
>
> Produce atomics, each carrying its source citation.

The framework distills atomics from the source; the user reviews and accepts or modifies before they land in the engram vault.

**For Creativity reference orientation**, the prompt is just:

> [Paste the Reference — Creativity from Knowledge and Values]
>
> Walk me through the framing. What does this say about how creativity works in the Ora system? What does the IIF refinement add?

The AI walks through the conceptual frame — generative search, value-filtering, action selection collapsed to recognition; what remains human; the empirical evidence path through the Spark Recognition Corpus.

The system as a whole is best built up incrementally — the MindSpec is a one-time substantial investment; the atomics accumulate over time; the Reference is read once and revisited when the system's purpose needs re-grounding. The compound interest of the substrate is what makes the system valuable; a thin substrate produces thin downstream work.

## What this system enables

What the Knowledge Production System does that the constituent components alone do not is produce **the substrate the entire rest of Ora reads**. Three things become possible with the system that are not possible with the parts in isolation:

- **Value-aware downstream work.** Without MindSpec, every framework that needs to know "does this fit the user's values" has to ask the user inline at decision time. With MindSpec, the values substrate is read directly — IIF's value-alignment math runs against it; MOM's Resolution Statement Objectivity Protocol Near-Miss Elicitation surfaces near-misses against the user's actual commitments; the Coordination Layer's escalation paths route to the user when value-laden decisions arise. The downstream work becomes value-aware without per-decision interrogation.
- **A traversable knowledge graph rather than a pile of notes.** Atomic notes alone are claims. Atomic notes with the 13-type relationship taxonomy are a knowledge graph that the meta-layer's semantic-similarity engine can traverse — a `bridges` link between two atomics from different sources surfaces a connection neither source named explicitly; a `contradicts` link surfaces a tension worth investigating; a `presupposes` link reveals a hidden assumption. The graph is what makes the substrate productive — a thousand atomics with no relationship typing is a pile; the same thousand with explicit links is a substrate the rest of the system can do real work with.
- **A conceptual story that makes the system's purpose legible.** The Reference is what lets the user understand what they are doing when they produce a MindSpec, when they produce atomics, when they make recognition decisions on IIF-surfaced candidates. Without the conceptual story, the components feel like disconnected tools. With the story, the components compose into a comprehensible apparatus — the MindSpec captures the values; the atomics build the knowledge substrate; IIF runs the combinatorial generation and value-filtering; recognition is the irreducible human contribution; the residue is the empirical evidence for what creativity is. The conceptual story is not separate from the apparatus; it is the apparatus's own self-description.

## Citations

The Knowledge Production System draws on several traditions in cognitive science, knowledge management, and creative cognition. The MindSpec Interview's structure draws on values-elicitation traditions in psychology (Schwartz Value Survey, Rokeach Values Survey) and clinical interview methodology (motivational interviewing, narrative therapy techniques) — the three-stage assessment combines forced-choice scenario instruments (high reliability) with open-ended life-context elicitation (high validity). The 66-entry library across 11 families is internal to Ora and emerged from cumulative cross-tradition synthesis (Buddhist commitments, Stoic commitments, contemporary virtue ethics, secular humanism); the universality-audit corrections (v0.2.2) recalibrated 26 defaults from framework-author residual bias to the general-population median.

The Knowledge Artifact Coach's atomic-note pattern draws on Niklas Luhmann's Zettelkasten methodology (one-claim-per-atomic; explicit typed links; emergence of structure through accumulated relationships) and on knowledge-graph literature (the 13-type relationship taxonomy is closer to RDF / OWL relationship typing than to free-text linking). The provenance hierarchy is internal to Ora and documented in the YAML schema (§5–§6.5); since rev 5.2 a kept engram weighs the same whether the user or the AI first typed it.

The Creativity reference draws on combinatorial creativity research (Boden, Koestler, Kauffman) for the generative-search framing; on Bayesian decision theory for the value-filtering framing; on Gestalt psychology and Daniel Kahneman's System 1 / System 2 distinction for the recognition-as-primitive framing. The IIF refinement (Creativity Collapses to Recognition; What Remains Human) is internal to Ora and was added 2026-05-08 alongside the IIF landing.

The framework versions composed here are MindSpec Interview (v0.2.3, single-file architecture restored 2026-05-09), KAC v6.0, and Reference — Creativity from Knowledge and Values (refined 2026-05-08). Per-framework citations will live in the per-framework papers when drafted.

## Open problems

- **The substrate's value is bounded by the user's labor, and nothing forces the labor.** MindSpec is a 90–120-minute investment; the engram corpus accumulates only through continuous coaching and document processing. A thin substrate produces thin downstream work, and this system — the most upstream of all — is the most dependent on effort it cannot compel.
- **MindSpec is a snapshot of values that drift silently.** It is updated through the Modifications log when the user notices a shift, but values change continuously and often unnoticed. A stale MindSpec misdirects every value-aware downstream decision until the user happens to update it.
- **Relationship typing is the load-bearing structural work, and it is a per-link judgment.** The 13-type taxonomy is what turns atomics from a pile into a traversable graph, but each link is the user's or the AI's call; a mis-typed or omitted link degrades the graph silently, and there is no check that the typing is correct.
- **Creativity-as-recognition is a hypothesis under measurement, not a settled result.** The Spark Recognition Corpus is the empirical apparatus, and the residue — recognitions the value model cannot predict — is the evidence being accumulated *for* the claim, not proof of it.