Skip to main content

vault-ingest-paper

Ingest, compare, or integrate an academic PDF into the External Brain vault with immutable source handling, page-level citations, atomic claims, and cross-paper links. Use for papers and PDF-based research, not ordinary Markdown articles.

Informations de source

Dépôt
Aznatkoiny/external-brain
Dernière activité de la source
20 août 2026 à 03:33
Langue détectée de SKILL.md
anglais
Étoiles
0
Forks
0

Options d'installation

Le prompt qui vérifie d'abord la source est sélectionné par défaut. Vous pouvez passer à une commande directe ou télécharger une copie locale.

Vérifiez les fichiers source

Lisez SKILL.md et les fichiers associés affichés par SkillsMP avant de décider de l'installer.

Explorateur de fichiers
2 fichiers

Affichage de SKILL.md

SKILL.md
Instructions source · Aperçu en lecture seule
name
vault-ingest-paper
description
Ingest, compare, or integrate an academic PDF into the External Brain vault with immutable source handling, page-level citations, atomic claims, and cross-paper links. Use for papers and PDF-based research, not ordinary Markdown articles.
# Vault Ingest Paper Compile an academic paper into source-grounded notes while preserving the original and the difference between reported findings and agent inference. ## Required context Read `vault/system/protocol.md`, `vault/system/schema.md`, `vault/system/relationships.md`, `templates/source-record.md`, and `skills/obsidian-markdown/SKILL.md`. Use the environment's PDF skill when available. ## Prepare the source 1. Ensure the original is under `vault/sources/papers/`. If the user supplied an attachment or a path outside the vault, preserve a byte-for-byte copy there first. Never overwrite an existing filename: reuse it when the SHA-256 matches, or use a distinct filename/version when the bytes differ. Do not alter it after ingestion. 2. Run `python3 scripts/extract_pdf.py "vault/sources/papers/Paper.pdf"`. Reuse the hash-addressed extraction if reported. 3. Read the extracted page-delimited Markdown. Render and inspect pages containing figures, tables, equations, unusual layouts, or suspicious extraction with `--render-pages`. 4. Verify title, authors, year, DOI, and page mapping from the paper. Leave unknown metadata empty. ## Integrate knowledge 1. Search the vault for the paper, its central concepts, methods, datasets, and claimed contributions. 2. Create one source record under `vault/sources/records/` from `templates/source-record.md`; store the PDF path and exact SHA-256. 3. Create or update only durable atomic claims, concepts, and questions. A source summary alone does not justify duplicating every section as a note. 4. Cite reported claims directly to PDF pages. Cite method, population, and experimental conditions when they qualify a result. 5. Compare against relevant existing claims. Describe `supports`, `challenges`, `extends`, or other relationships with their basis and citations. 6. Mark cross-source reasoning not stated by a source as `evidence_kind: agent-inference` and `status: agent-proposed`. 7. Put possible contradictions in `vault/system/review-queue.md` as `potential-conflict`; do not resolve them autonomously. 8. Update `vault/maps/`, `vault/system/index.md`, and `vault/system/log.md`, then run `python3 scripts/vault_lint.py`. For batch ingestion, process one paper to a coherent diff at a time so provenance and revisions remain reviewable.
Voir sur GitHub