feat(markdown-parser): support Obsidian wiki-link extraction ([[target|display]])

[WesleyMFrederick]

Problem

Citation-manager cannot extract content from Obsidian wiki-links ([[target|display]], [[target#header|display]], [[target#^block-id|display]]). When reading a document containing wiki-links with %%force-extract%% tags, the citation-extractor hook returns zero extracted content because the markdown parser doesn't recognize wiki-link syntax as links.

This breaks the progressive-disclosure workflow: design documents that reference PRDs and whiteboards via wiki-links get no auto-extracted context.

Reproduction Steps

1. Create a markdown file with wiki-links:

   **Inputs**: [[continuous-learning-port-prd|PRD]]%%force-extract%%

2. Run: citation-manager extract links <file>
3. Observe output:

   {
     "extractedContentBlocks": { "_totalContentCharacterLength": 2 },
     "outgoingLinksReport": { "processedLinks": [] },
     "stats": { "totalLinks": 0 }
   }

4. Expected: PRD content extracted. Actual: 0 links found.

Root Cause

The markdown parser (likely marked) does not recognize Obsidian wiki-link syntax ([[...]]) as link tokens. Wiki-links appear as plain text in the AST, so the content extractor never processes them.

The AST output confirms wiki-links are embedded inside paragraph → text tokens with no link-type node generated.

Expected Behavior

• [[target|display]] resolved to a file path relative to the document's directory
• [[target#header|display]] resolved to file + header section extraction
• [[target#^block-id|display]] resolved to file + block anchor extraction
• %%force-extract%% tag works with wiki-links (extracts full file content)
• Citation-validator hook auto-extracts linked content when reading .md files

Related

• Citation validator false negative: Internal links with backticks in anchor IDs cc-workflows#27 (citation validator false negative — related parser gap)
• Design doc that exposed this: /WesleyMFrederick/cc-workflows/blob/main/design-docs/features/20260202-continuous-learning-port/continous-learning-port-design.md
• All Obsidian-native vault documents use wiki-links as primary linking format

Note

This is a high priority issue because the entire vault uses Obsidian wiki-links. Without this, citation-extractor provides zero value for design docs, PRDs, and whiteboards that cross-reference each other. The %%force-extract%% workflow is broken for the primary link format in use.

Wiki-link resolution rules (Obsidian):

• [[filename]] — resolves to closest matching .md file by name (no path needed)
• [[filename|display]] — same resolution, display text differs
• [[filename#Header]] — file + header section
• [[filename#^block-id]] — file + block anchor
• Resolution is relative to vault root, with shortest-path matching

---

Acceptance Criteria

• Parser detects [[target]] wiki-link syntax in markdown content
• Parser detects [[target|display]] with display text
• Parser detects [[target#header]] with header references
• Parser detects [[target#^block-id]] with block anchor references
• Wiki-links resolve to actual file paths relative to document location
• %%force-extract%% tag works with wiki-links (full file extraction)
• citation-manager extract links returns extracted content from wiki-linked files
• Citation-extractor hook fires correctly for wiki-link references
• Existing markdown link extraction ([text](path)) unchanged (no regression)

Definition of Done

• Failing tests written (RED phase)
• Implementation complete (GREEN phase)
• All existing tests pass (no regression)
• New tests cover all wiki-link variants
• Build succeeds
• Committed with conventional commit

Component Guides

• MarkdownParser Component Guide
• ContentExtractor Component Guide