Knowledge Packs — Stunspot’s Guide to Human Behavioral Neurobiology
The repository ships the same canon in three upload shapes: individual reports, compiled packs, and an omnibus file. The content is intentionally redundant across formats. Pick one shape for your workflow unless your retrieval system has explicit deduplication.
Recommended Default
Use the compiled packs in knowledge-packs/compiled-packs/ for most AI Projects, RAG systems, long-context workspaces, and NotebookLM-style tools.
They give the model the whole canon in four coherent files while preserving the volume structure. It is the best balance of coverage, orientation, and file-count sanity.
Pack Types
| Pack Type | Files | Path | Best Use |
|---|---|---|---|
| Source reports | 12 | knowledge-packs/by-report/ |
Precise retrieval, selective upload, report-level citation, corpus inspection, editing. |
| Compiled packs | 4 | knowledge-packs/compiled-packs/ |
Recommended default: grouped coverage with lower file count and preserved volume logic. |
| Omnibus | 1 | knowledge-packs/omnibus/ |
One-file import, local archive, whole-corpus indexing, or strong long-context systems. |
Decision Guide
Use source reports when:
- you need citations tied to a specific report;
- you are building a high-quality RAG index with metadata filters;
- you want to upload only part of the canon;
- you are editing, auditing, or sampling the corpus.
Use compiled packs when:
- you want the canon to work immediately inside an AI project;
- your tool has a moderate file-count limit;
- you want the model to retain volume-level structure;
- you need broad coverage without juggling 12 files.
Use the omnibus when:
- your tool prefers one file;
- you want an archival copy;
- you are using a long-context model;
- you are building a local search or indexing workflow that handles large Markdown files cleanly.
Avoid mixing pack types in the same naive retrieval index. Loading by-report/, compiled-packs/, and omnibus/ together can cause duplicate retrieval and overweight repeated passages.
Source Reports
Source reports are the canonical individual units. Use them for precise retrieval, selective upload, source-level citation, or manual inspection.
Compiled Packs
Compiled packs are the recommended default upload shape.
| Pack | Coverage | Path |
|---|---|---|
| Volume 1. A-D Foundations of Adaptive Regulation | Foundational ontology, evidence discipline, predictive/allostatic engine, metabolic ledger | knowledge-packs/compiled-packs/knowledge-volume-1-a-d-foundations-of-adaptive-regulation.md |
| Volume 2. E-G Major Operating Domains I | Chemical tuning, stress/sleep/immunity, behavioral ecology, affordance landscapes | knowledge-packs/compiled-packs/knowledge-volume-2-e-g-major-operating-domains-i.md |
| Volume 3. H-J Major Operating Domains II | Social regulation, culture/narrative/ideology, digital attention ecologies | knowledge-packs/compiled-packs/knowledge-volume-3-h-j-major-operating-domains-ii.md |
| Volume 4. K-L Constraint, Breakdown, and Diagnostic Layers | Feedback loops, systemic failure modes, state-trait differentiation, diagnostic guardrails | knowledge-packs/compiled-packs/knowledge-volume-4-k-l-constraint-breakdown-and-diagnostic-layers.md |
Omnibus
The omnibus is the entire canon in one file:
Use it for archival import, local search, or long-context systems. Avoid it in small RAG systems that chunk poorly or lose heading metadata.
Suggested RAG Metadata
canon: "Stunspot's Guide to Human Behavioral Neurobiology"
version: "1.0"
release_date: "2026-06-28"
license: "CC BY-NC-SA 4.0"
author: "Sam “stunspot” Walker / Collaborative Dynamics"
pack_type: "source_report | compiled_pack | omnibus"
report_code: "A-L if applicable"
volume: "1-4 if applicable"
repo_path: "knowledge-packs/..."
Practical Upload Recommendations
| Workflow | Recommended Upload |
|---|---|
| ChatGPT Project / Claude Project / Gemini Gem-style knowledge | 4 compiled packs |
| NotebookLM-style broad study | 4 compiled packs or omnibus |
| RAG index with citation control | 12 source reports |
| Local archive | Omnibus plus manifest files |
| Focused analysis on one domain | Relevant source reports only |
| Model instruction/persona development | Compiled packs plus Report L if using selective upload |