Source ingest web article

source-ingest-web-article

Ingest exactly one web article source into an Obsidian-style Markdown knowledge base or personal wiki. Use this for Markdown article captures, clipped web pages, newsletter exports, and article-like source files. Do not use it for books, long PDFs, podcasts, videos, or multi-source batch ingestion.

Operating Principles

• Process exactly one source at a time.
• Follow local knowledge-base instructions first, including existing vault rules, naming conventions, metadata schemas, citation style, asset handling, index policies, and processed-source workflow.
• Prefer the source path provided by the user.
• If no path is provided, look for an obvious inbox, source, clipping, import, or article folder. If there is no clear single source, ask the user for the article source path.
• Treat raw sources as immutable unless local rules explicitly allow safe metadata or asset-link updates.
• Use any locally available Markdown, Obsidian, retrieval, search, or domain-specific capabilities when relevant and discoverable from the environment. Do not assume private tool names.
• Keep citations close to claims. Prefer concise paraphrase over long quotes.
• Preserve specific claims, numbers, dates, evidence, uncertainty, and source context.

Workflow

1. Identify local rules.

   • Read project or vault instructions before editing notes.
   • Detect local conventions for frontmatter, tags, folders, note titles, citations, indexes, source catalogs, logs, processed folders, and assets.
   • Check whether a search or retrieval index must be refreshed after changes.

2. Select the source.

   • Use the user-provided source path when available.
   • If absent, find one obvious unprocessed article source.
   • If multiple plausible sources exist, stop and ask the user which one to ingest.
   • Confirm the source is article-like. If it is a book, long PDF, podcast, video, or batch folder, explain that this skill is not the right workflow.

3. Inspect the article.

   • Extract title, author, publication, original URL, publication date, clipped date, source file path, and capture metadata when present.
   • Read for important claims, entities, concepts, dates, evidence, examples, uncertainties, contradictions, and open questions.
   • Note article scope and limitations.

4. Create or update a source summary note.

   • Reuse an existing source summary if one clearly corresponds to the article.
   • Otherwise create one using local naming and folder conventions.
   • Use internal wiki links for the first meaningful mention of important people, organizations, places, concepts, questions, and sources.
   • Do not create links for every named thing; prioritize entities and concepts that matter to the article or the knowledge base.

5. Update topic, entity, and concept notes selectively.

   • Update an existing note when the article adds a new durable claim, new evidence, a correction, a contradiction, or a useful example.
   • Do not update a note merely because the article mentions the topic in passing.
   • Create new notes only for recurring or likely-to-recur concepts, entities, places, questions, or sources.
   • Keep additions concise, clearly attributed, and consistent with local metadata conventions.
   • If a note is updated, add the article to source or reference metadata when appropriate and update its modified date if the knowledge base uses one.

6. Check contradictions and uncertainty.

   • Search existing notes for claims that conflict with the article.
   • If there is a conflict, update the relevant contradictions, uncertainty, or open-questions note if one exists.
   • Link the conflicting notes and the source summary.
   • State the conflicting claims, source basis, and current resolution status.
   • If resolution is unclear, mark uncertainty instead of forcing a conclusion.

7. Update indexes and logs.

   • Update index, map-of-content, dashboard, source catalog, or recently updated notes whenever new notes are created or substantially changed, if the knowledge base has them.
   • Append an ingest entry to a log or changelog note if one exists.
   • Keep index and log changes factual and compact.

8. Finish source handling.

   • Move or mark the source as processed only after the ingest is complete and local rules allow it.
   • If blocked or incomplete, leave the source in its original location, mark the summary as needing review, and clearly explain what remains.

Source Summary Structure

Use local frontmatter conventions when they exist. If none exist, prefer this portable structure:

---
categories:
  - sources
tags:
  - source-ingest
created: YYYY-MM-DD
updated: YYYY-MM-DD
sources: []
status: active
---

Recommended sections:

# Source Title

## Summary

## Key Claims

## Evidence and Examples

## Important Entities

## Concepts and Links

## Dates and Timeline

## Open Questions

## Possible Contradictions

## Source Notes

For source notes:

• Include original URL, clipped source path, author, publication, publication date, capture date, and access date when available.
• Cite the raw source file, original URL, clipped article, or source note according to local convention.
• Keep citations close to the claims they support.
• Use short quotes only when wording matters.

Image and Asset Workflow

1. Detect images.

   • Find remote Markdown images, reference-style Markdown images, and HTML img tags.
   • Identify meaningful article images, charts, screenshots, diagrams, and figures.
   • Skip decorative, tracking, spacer, avatar, logo-only, and social-share images unless central to the article.

2. Localize assets only if allowed.

   • Download meaningful article images into a local asset folder associated with the source when local rules allow asset localization.
   • Follow local asset folder naming, relative-link style, and attachment conventions.
   • Leave non-image embeds, such as YouTube or social embeds, remote.

3. Optimize images when appropriate.

   • Prefer WebP or another locally preferred efficient format when it meaningfully reduces size.
   • Cap very wide raster images unless detail would be lost.
   • Use quality settings that preserve chart labels, screenshots, and text readability.
   • Keep SVGs as-is.
   • Keep animated GIFs only when motion matters; otherwise use a representative still image.

4. Rewrite embeds carefully.

   • Rewrite image embeds to local relative links only when the source format and local rules allow it.
   • Do not rewrite raw source files unless local rules explicitly permit it.
   • If image download or optimization fails, continue the ingest and record blocked downloads in the ingest log or final summary.

Topic Note Decision Rules

Update a note when:

• The article adds a durable claim the note should remember.
• It provides new evidence, a useful example, a correction, or a contradiction.
• It clarifies a timeline, definition, relationship, or uncertainty.
• It materially changes how the knowledge base should understand the topic.

Do not update a note when:

• The article only mentions the topic in passing.
• The information is redundant with an existing well-cited claim.
• The entity is incidental and unlikely to recur.
• The update would create noisy one-off backlinks without adding knowledge.

Contradiction Entry Example

## [YYYY-MM-DD] Source Title

- Conflict: [[Existing Note]] says ..., while [[Source Title]] says ...
- Basis: Existing note cites ...; new source cites ...
- Status: unresolved
- Follow-up: ...

Ingest Log Entry Example

## [YYYY-MM-DD] ingest | Source Title

- Source: ...
- Summary: ...
- Notes created:
  - ...
- Notes updated:
  - ...
- Assets:
  - ...
- Follow-up:
  - ...

Validation Checklist

• Exactly one source was processed.
• Local project instructions were followed.
• Raw source was not rewritten except where local rules allow it.
• Source summary exists and has valid frontmatter if frontmatter is used.
• Important internal links were added.
• Relevant concept, entity, topic, or question notes were updated.
• New notes follow local metadata conventions.
• Index, source catalog, dashboard, or log notes were updated if applicable.
• Contradictions, uncertainties, and open questions were recorded when applicable.
• Article images were handled according to local rules, with blocked downloads noted.
• Source was moved or marked processed only if ingest is complete and local rules allow it.
• Search or retrieval index was refreshed if the project has one.
• Final response lists changed files and any blocked items.