Back to corpus
proposalexperiment writeup candidatescore 34

cc-graph-kernel

The Graph Kernel transforms a 107K+ turn conversation DAG into deterministic, replayable context slices. It answers one question:

Open full HTMLDownload HTMLCorpus JSONPaper schema

Full HTML reader

Read the full artifact

Open in new tab

Extracted abstract or opening context

The Graph Kernel transforms a 107K+ turn conversation DAG into deterministic, replayable context slices. It answers one question: 1. [Overview](#overview) 2. [Core Contract](#core-contract) 3. [Architecture](#architecture) 4. [Installation](#installation) 5. [Quick Start](#quick-start) 6. [Types Reference](#types-reference) 7. [SlicePolicy v1](#slicepolicy-v1) 8. [Priority Scoring](#priority-scoring) 9. [Context Slicer Algorithm](#context-slicer-algorithm) 10. [Graph Stores](#graph-stores) 11. [SliceExport & Fingerprinting](#sliceexport--fingerprinting) 12. [Determinism Guarantees](#determinism-guarantees) 13. [Integration with CognitiveTwin Bridge](#integration-with-cognitivetwin-bridge) 14. [Testing](#testing) 15. [API Reference](#api-reference) The Graph Kernel is the "context construction engine" for the semantic kernel stack. It sits between raw conversation data and semantic analysis, providing: - **Bounded context windows** around any anchor turn - **Phase-aware prioritization** (Synthesis > Planning > Consolidation > Debugging > Exploration) - **Deterministic fingerprints** for reproducibility and provenance - **Salience-weighted expansion** to capture the most important turns first Without deterministic context slicing: - "Same input" can produce different semantic analyses - Provenance becomes meaningless - Replay and debugging are impossible - Evidence accumulation can't be trusted

Promotion decision

What has to happen next

Attach run IDs, datasets, metrics, and reproduction commands.

Why this is not always a full paper yet

Corpus pages are public-safe readers for discovered workspace artifacts. They are not automatically final papers. A corpus item becomes a polished paper only after the editable source, evidence checkpoints, references, figures, render path, and release status are attached through the paper schema.