Sources

External sources snapshotted into the KB — articles, papers, GitHub issues, tweets, repository READMEs — along with their ingest reports.

Each source is captured verbatim into a file at the time of snapshot and analyzed in a paired *.ingest.md report that classifies it, extracts claims, and links it into the rest of the KB.

What's here

• Source snapshots — one file per snapshotted external artifact, with a short header noting the URL and capture date followed by the verbatim content.
• *.ingest.md reports — the analyses produced by cp-skill-ingest. Each report names the source, classifies it as a related system / paper / article / etc., summarises the load-bearing claims, and proposes connections into kb/notes/ and other collections.
• types/ — local type definitions for the artifacts in this collection (source-review, ingest-report).
• dir-index.md — auto-generated alphabetical listing of every file in this collection.

How to add a source

# URL → snapshot + ingest in one go
cp-skill-ingest https://example.com/some-article

# Or snapshot first, ingest later
cp-skill-snapshot-web https://example.com/some-article
cp-skill-ingest kb/sources/some-article.md

The skill chooses the right snapshotting backend based on the URL (GitHub API for issues/PRs, the X SDK for tweets, plain WebFetch for everything else).

Collection conventions

This collection operates in the descriptive register — sources are captured for fidelity, not transformed. Writing conventions for the *.ingest.md reports live in types/ingest-report.md. For the discovery and connection conventions used during ingest, see the cp-skill-ingest skill.