LOGOS Source Model v0
LOGOS is the NEXUS concept area for intake, refinement, and retention of knowledge-bearing signals.
This document defines the first small structural model for that area.
Purpose
The goal of this first pass is not to define the whole LOGOS ontology.
The goal is to name the smallest stable concepts needed to represent where a signal came from and what kind of signal it is, so future intake paths can converge without losing meaning.
This is especially relevant for later sources such as:
- AI conversations
- published blog posts and articles
- forum threads
- Talkyard forum threads
- Discord channels and threads
- email threads
- bug reports
- support requests
- deployed-app feedback
Stable Concepts
The v0 LOGOS source model introduces:
SourceSystemIdthe stable slug for the originating system or source surfaceSourceInstanceIdthe stable slug for one concrete source authority or deployment within that source systemIntakeChannelIdthe stable slug for the intake channel or pathSignalKindIdthe stable slug for the kind of knowledge-bearing signalAccessContextIdthe stable slug for the visibility or authority context under which the material was observedAcquisitionKindIdthe stable slug for how the material entered NEXUSRightsPolicyIdthe stable slug for the reuse boundary governing the materialLogosLocatora concrete locator back to the originating source itemLogosSourceRefthe source-system + intake-channel + locator groupingLogosSignala small semantic envelope for a captured signalLogosHandlingPolicya small explicit handling envelope covering sensitivity, sharing scope, sanitization status, and retention classLogosAccessContextthe explicit source-instance + access-context + acquisition-kind envelopeLogosRightsContextthe explicit rights-policy + attribution-reference envelope- pool boundary types
explicit
raw,private, andpublic-safehandling boundaries for downstream use
Relation To Existing NEXUS Layers
This model does not replace canonical ingestion history.
Instead:
- canonical history still records what was observed or imported
- LOGOS source types classify the source and meaning of knowledge-bearing intake
- handling policy describes the material
- pool boundary types constrain what downstream workflows may do with it
- provider and Codex imports can now carry LOGOS metadata from the moment they enter NEXUS
- later curation, refinement, retrieval, and doctrine can build on both
In other words:
- canonical history answers: what did NEXUS observe?
- LOGOS source modeling answers: what kind of source or signal is this?
Overlap
LOGOS must expect source overlap.
The same underlying interaction may later be seen through multiple acquisition paths:
- local session capture
- provider export
- API capture
- copied or quoted text in another system
That overlap should be preserved separately first and reconciled explicitly later.
See:
v0 Scope
The v0 scope is intentionally narrow:
- stable source-system identifiers
- stable intake-channel identifiers
- stable signal-kind identifiers
- minimal source references and signal envelopes
- explicit access and rights metadata for manual LOGOS intake notes
- a small durable note workflow for seeding non-chat intake before full ingestion exists
- a first explicit derived-note workflow for redacted, anonymized, or shareable LOGOS derivatives
- a first handling-policy audit report over LOGOS note material
- first explicit pool boundary types for
raw,private, andpublic-safeuse, with public-safe export now also constrained by rights policy - explicit pool-aware intake and derived note paths under
docs/logos-intake/<pool>/anddocs/logos-intake-derived/<pool>/ - provider and Codex imports entering the system with restricted-by-default LOGOS handling metadata
- attribution requirements surfaced into public export manifests
Deferred:
- full LOGOS ontology
- vector or embedding storage
- retrieval ranking
- clustering
- doctrine/refinement workflows
- broader redaction and anonymization workflows beyond the first derived-note path
- cross-source overlap reconciliation implementation
Design Notes
v0 LOGOS identifiers use explicit allowlisted slug validation rather than permissive acceptance.
That keeps this layer aligned with the repo rule that stable recognized forms should be narrow and deterministic.
Current concrete source-system allowlist:
chatgptclaudegrokcodexblogforumtalkyarddiscordemailissue-trackerapp-feedback-surface
Current concrete intake-channel allowlist:
ai-conversationpublished-articleforum-threaddiscord-channeldiscord-threademail-threadbug-reportapp-feedback
Current concrete signal-kind allowlist:
conversationmessagearticlebug-reportfeedbacksupport-question
Current handling-policy allowlists:
- sensitivities:
personal-privatecustomer-confidentialinternal-restrictedpublic- sharing scopes:
owner-onlycase-teamproject-teampublic- sanitization statuses:
rawredactedanonymizedapproved-for-sharing- retention classes:
ephemeralcase-bounddurable
Current access and rights allowlists:
- access contexts:
public-anonymousregistered-userowneradminbotapi-client- acquisition kinds:
manual-notegit-syncweb-scrapeapi-pullmanual-exportlive-capture- rights policies:
owner-controlledpersonal-training-onlysite-terms-restrictedcc-bycc-by-saapi-contract-restrictedcustomer-confidentialreview-required
Current provider-import baseline:
- provider and Codex imports enter with restricted-by-default handling metadata
- current default handling is:
sensitivity = internal-restrictedsharing_scope = owner-onlysanitization_status = rawretention_class = durableentry_pool = raw
Current non-chat LOGOS note baseline:
- manual intake notes now enter a declared pool at creation time
- manual intake notes now also persist explicit source-instance, access, acquisition, rights, and optional attribution-reference metadata
- the default entry pool is:
entry_pool = raw- the default access/rights posture is:
access_context = owneracquisition_kind = manual-noterights_policy = review-required- derived sanitized notes resolve into:
privatewhen the resulting policy is still restrictedpublic-safeonly when the explicit public-safe policy boundary is crossed for both handling and rights
Published writing note:
- owner-controlled public Markdown writing can now enter LOGOS through a Git-backed import path
- the current concrete form is a Markdown blog repository imported as:
source_system = blogintake_channel = published-articlesignal_kind = articleacquisition_kind = git-sync- those notes are modeled as
public-safefrom entry only when the rights and handling policy explicitly allow that
Discord note:
- Discord is modeled as a LOGOS source now so later inbound ingestion and later outbound NEXUS-to-Discord flows can share the same source vocabulary.
- The current scope is intake classification only, not outbound transport or live Discord integration.
Talkyard note:
forumremains the generic category.talkyardis now the explicit current forum implementation source system.- That lets NEXUS model real Talkyard intake now without losing the broader forum abstraction later.