Knowledge Packs — Stunspot’s Guide to Investigative News Intelligence
The knowledge-packs/ directory contains the model-facing corpus. Use these files when loading the canon into AI Projects, RAG systems, long-context workspaces, agent memory, local search, or research notebooks.
The docs/ directory is only navigation and usage guidance. The actual report corpus lives here.
Recommended Default
Use the compiled packs unless you have a specific reason not to.
They preserve the canon’s volume structure, keep related concepts together, and reduce upload friction for tools that perform poorly with many separate files. The source-report directory also contains the individual A–M reports for precise report-level retrieval and editing.
Pack Types
| Pack | Files | Location | Best Use |
|---|---|---|---|
| Source reports | 13 | knowledge-packs/by-report/ |
Precise retrieval, selective upload, report-level citation, targeted editing. |
| Compiled packs | 4 | knowledge-packs/compiled-packs/ |
Recommended default: grouped coverage with lower file count and full A–M sequence. |
| Omnibus | 1 | knowledge-packs/omnibus/ |
One-file import, local archive, whole-canon search, or strong long-context/RAG systems. |
Selection Guide
Choose source reports when:
- your RAG system needs narrow provenance and report-level citation;
- you are editing or auditing a specific report;
- you want to load only one conceptual area, such as signal judgment or information operations;
- your tool chunks each file cleanly and preserves headings in metadata.
Choose compiled packs when:
- you are uploading the canon into an AI Project or NotebookLM-style workspace;
- your tool has file-count limits;
- you want the model to see each volume as a coherent conceptual unit;
- you need the full A–M sequence without loading one large omnibus file.
Choose the omnibus when:
- your system handles large files well;
- you want one canonical archive artifact;
- you are doing local search, grep-style lookup, or whole-canon embedding;
- your long-context model benefits from seeing the entire sequence in one contiguous document.
Source Reports
The source reports are the canonical individual A–M units.
- A. Reality Models, Event Ontology & Public-Reality Mapping.md — ground reality, public reality, event ontology, traces, evidence, silence, uncertainty, confidence, and reality projection.
- B. Signal, Evidence & Confidence Architecture.md — signal as relational value, the judgment stack, evidence classes, confidence ceilings, corroboration, and false convergence.
- C. Information Ecosystems, Source Topologies & Platform Behavior.md — source topology, routing, origin nodes, bridge nodes, platform affordances, amplification, stabilization, decay, and resurfacing.
- D. Topic Discovery, Weak-Signal Detection & Story Emergence.md — weak signals, anomaly detection, pre-narrative coherence, latent story objects, escalation thresholds, maturation windows, and kill criteria.
- E. Narrative Divergence, Coverage Gaps & Missingness Analysis.md — comparative rendering, expected-versus-observed visibility, missingness, strategic silence, selective omission, and divergence analysis.
- F. Thread Following, Link Expansion, and Investigative Graph Construction.md — entrypoint nodes, expansion operators, typed relations, directional asymmetry, edge status, entity resolution, and graph integrity.
- G. Temporal Dynamics, Trend Formation, and Story Evolution.md — four investigative clocks, trigger events, substrate development, temporal regimes, phase transitions, narrative pivots, and delayed confirmation.
- H. Power, Incentives & Strategic Actor Analysis.md — actors as strategic nodes, incentives, leverage, dependency, influence, control, alignment, coordination, and power mapping.
- I. Propaganda, Narrative Laundering & Information Operations.md — operation analysis, narrative seeding/routing/uptake, credibility laundering, false plurality, agenda manipulation, and memetic weaponization.
- J. Implied Reality, Causal Inference, and Unstated-Truth Extraction.md — causal pressure, bounded inference, hidden knowledge states, hidden dependencies, second-order implications, contradiction resolution, and null nodes.
- K. Editorial Prioritization, Escalation Logic & Intervention Strategy.md — editorial importance, scarce-attention governance, lead quality, escalation logic, intervention choices, and kill criteria.
- L. Failure Modes, Hallucinated Patterns & Analytical Breakdown.md — repetition versus corroboration, salience versus significance, false coherence, analytical contamination, and failure diagnosis.
- M. Analyst Workflows, Research Artifacts & Publication Systems.md — workflow architecture, durable research artifacts, handoffs, publication systems, and execution discipline.
Compiled Packs
The compiled packs are the recommended operational upload set.
- [KNOWLEDGE] - Investigative News Intelligence - Vol. 1 A-C Foundations of Reality, Signal, and Judgment.md
- [KNOWLEDGE] - Investigative News Intelligence - Vol. 2 D-G Discovery and Thread Expansion Domains.md
- [KNOWLEDGE] - Investigative News Intelligence - Vol. 3 H-J Interpretation and Adversarial Sensemaking.md
- [KNOWLEDGE] - Investigative News Intelligence - Vol. 4 K-M Editorial Control, Failure Diagnosis, and Execution.md
Omnibus
Use the omnibus when continuity matters more than modularity.