docs(extraction): link fixes and remove stale release version tags (NVBug 6179241)

[kheiss-uwzoo]

Summary

Retargets GitHub source-code deep links in extraction docs after the NeMo Retriever module reorg. The crawl failures were caused by moved .py paths in the repository, not by incorrect .md page content or nav structure.

Root cause: Python modules moved from legacy paths (vdb/, service/models/, evaluation/) to their current locations (common/vdb/, common/schemas/, operators/, tools/evaluation/). Doc hrefs were not updated when the code moved.

Scope (path-only, no prose/nav changes):

Page	Updates
vdbs.md	vdb/lancedb.py, vdb/adt_vdb.py → common/vdb/; vdb/operators.py → operators/vdb.py; tree links → common/vdb
custom-metadata.md	service/models/*.py → common/schemas/; VDB tree links → common/vdb
workflow-document-ingestion.md	evaluation/README.md → tools/evaluation/README.md (same anchor)

Out of scope (handled elsewhere): custom-metadata page consolidation (#2195 / #2203), MkDocs anchors (#2245), support-matrix edits.

Fixes the 10 broken GitHub deep links on the latest extraction track (NVBug 6179241). 26.3.0 was already resolved in #2249 / #2250.

Test plan

• Verified all updated GitHub blob/tree URLs return HTTP 200 on main
• After merge + docs publish: re-crawl docs.nvidia.com/nemo/retriever/latest/extraction/ — expect 0 broken content links on these three pages
• If remove custom-metadata.md; canonical path is vdbs + notebooks #2195 or docs(extraction): sync post-#2179 extraction doc fixes to main #2203 merges first and removes custom-metadata.md, cherry-pick or rebase this commit onto the consolidated vdbs.md paths only

[greptile-apps]

Greptile Summary

This PR removes stale release-version tags (26.05) from extraction docs and Helm READMEs, and retargets blob/26.05 GitHub deep links to blob/main. It also renames the #image-captioning-2605 anchor to #image-captioning across all referencing files for consistency.

• Link/anchor updates (deployment-options.md, prerequisites-support-matrix.md, helm/README.md, openshift.md): All blob/26.05 GitHub URLs are replaced with blob/main; the #image-captioning-2605 fragment is renamed to #image-captioning everywhere it appears, with the heading anchor in prerequisites-support-matrix.md updated to match.
• Content removal (workflow-document-ingestion.md, nemo_retriever/README.md): The stale CLI ingestion section (referencing old evaluation paths) is removed from the extraction doc; the agentic retrieval section is removed from the library README.
• Missed fix (nemo_retriever/README.md): Two links at lines 150 and 169 still point to ./src/nemo_retriever/evaluation/README.md (old path); the file now lives at ./src/nemo_retriever/tools/evaluation/README.md.

Confidence Score: 4/5

Safe to merge with one outstanding broken link in the library README that this PR was positioned to fix.

Six of the seven changed files are clean documentation link and prose updates. The one gap is in nemo_retriever/README.md: the 'Ingest a test corpus (CLI)' section at lines 150 and 169 still links to ./src/nemo_retriever/evaluation/README.md, a path that no longer exists — the file moved to ./src/nemo_retriever/tools/evaluation/README.md. The PR removes a parallel broken reference in workflow-document-ingestion.md but leaves these two intact.

nemo_retriever/README.md — lines 150 and 169 contain broken relative links to the evaluation README that need to be updated to the current path.

Important Files Changed

Filename	Overview
nemo_retriever/README.md	Removes the 'Run agentic retrieval' section; retains the 'Ingest a test corpus (CLI)' section which still has two broken links to the old evaluation README path (./src/nemo_retriever/evaluation/README.md at lines 150 and 169).
docs/docs/extraction/workflow-document-ingestion.md	Removes the 'Ingest a test corpus (CLI)' section (which contained the stale evaluation README link) and trims a sentence referencing the graph_pipeline CLI. Clean removal with no broken anchors left behind.
docs/docs/extraction/prerequisites-support-matrix.md	Renames heading from 'Image captioning (26.05) { #image-captioning-2605 }' to 'Image captioning { #image-captioning }' and updates all branch-pinned links from blob/26.05 to blob/main. Anchor rename is consistent with updates in openshift.md and helm/README.md.
docs/docs/extraction/deployment-options.md	Replaces blob/26.05 GitHub links with blob/main, removes inline '26.05' version tags from prose, and updates the air-gapped section anchor from #image-captioning-2605 to #image-captioning to match the renamed heading in prerequisites-support-matrix.md.
docs/docs/extraction/concepts.md	Pure prose edits: 'see' → 'refer to' throughout. No link or anchor changes; no issues.
nemo_retriever/helm/README.md	Removes '26.05' version pins from prose and table descriptions; updates blob/26.05 GitHub links to blob/main; changes #image-captioning-2605 anchor references to #image-captioning. The { #recommended-minimal-install-2605 } anchor on the heading is preserved, so all inbound links to that fragment remain valid.
nemo_retriever/helm/openshift.md	Updates one anchor fragment from #image-captioning-2605 to #image-captioning, consistent with the heading rename in prerequisites-support-matrix.md. No other changes.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[PR: Remove stale 26.05 version tags] --> B[docs/docs/extraction/]
    A --> C[nemo_retriever/helm/]
    A --> D[nemo_retriever/README.md]

    B --> B1[concepts.md\nsee → refer to prose]
    B --> B2[deployment-options.md\nblob/26.05 → blob/main\nanchor: image-captioning-2605 → image-captioning]
    B --> B3[prerequisites-support-matrix.md\nHeading renamed: image-captioning-2605 → image-captioning\nblob/26.05 → blob/main]
    B --> B4[workflow-document-ingestion.md\nRemoved CLI section with stale evaluation link]

    C --> C1[helm/README.md\n26.05 prose removed\nblob/26.05 → blob/main\nanchor refs updated]
    C --> C2[helm/openshift.md\nanchor: image-captioning-2605 → image-captioning]

    D --> D1[Removed agentic retrieval section]
    D --> D2[Lines 150 and 169 still link to\nsrc/nemo_retriever/evaluation/README.md\nShould be: tools/evaluation/README.md]

    style D2 fill:#ff6b6b,color:#fff

Loading

%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
flowchart TD
    A[PR: Remove stale 26.05 version tags] --> B[docs/docs/extraction/]
    A --> C[nemo_retriever/helm/]
    A --> D[nemo_retriever/README.md]

    B --> B1[concepts.md\nsee → refer to prose]
    B --> B2[deployment-options.md\nblob/26.05 → blob/main\nanchor: image-captioning-2605 → image-captioning]
    B --> B3[prerequisites-support-matrix.md\nHeading renamed: image-captioning-2605 → image-captioning\nblob/26.05 → blob/main]
    B --> B4[workflow-document-ingestion.md\nRemoved CLI section with stale evaluation link]

    C --> C1[helm/README.md\n26.05 prose removed\nblob/26.05 → blob/main\nanchor refs updated]
    C --> C2[helm/openshift.md\nanchor: image-captioning-2605 → image-captioning]

    D --> D1[Removed agentic retrieval section]
    D --> D2[Lines 150 and 169 still link to\nsrc/nemo_retriever/evaluation/README.md\nShould be: tools/evaluation/README.md]

    style D2 fill:#ff6b6b,color:#fff

Loading

Comments Outside Diff (1)

1. nemo_retriever/README.md, line 150-169 (link)

   Broken evaluation README links not fixed

   Lines 150 and 169 both link to ./src/nemo_retriever/evaluation/README.md, but the file now lives at ./src/nemo_retriever/tools/evaluation/README.md — confirmed on disk. This PR fixes the matching link in docs/docs/extraction/workflow-document-ingestion.md by removing the section entirely, but these two links in the library README remain broken and will 404 for anyone following them.

   Prompt To Fix With AI

   This is a comment left during a code review.
   Path: nemo_retriever/README.md
   Line: 150-169
   
   Comment:
   **Broken evaluation README links not fixed**
   
   Lines 150 and 169 both link to `./src/nemo_retriever/evaluation/README.md`, but the file now lives at `./src/nemo_retriever/tools/evaluation/README.md` — confirmed on disk. This PR fixes the matching link in `docs/docs/extraction/workflow-document-ingestion.md` by removing the section entirely, but these two links in the library README remain broken and will 404 for anyone following them.
   
   How can I resolve this? If you propose a fix, please make it concise.

Prompt To Fix All With AI

Fix the following 1 code review issue. Work through them one at a time, proposing concise fixes.

---

### Issue 1 of 1
nemo_retriever/README.md:150-169
**Broken evaluation README links not fixed**

Lines 150 and 169 both link to `./src/nemo_retriever/evaluation/README.md`, but the file now lives at `./src/nemo_retriever/tools/evaluation/README.md` — confirmed on disk. This PR fixes the matching link in `docs/docs/extraction/workflow-document-ingestion.md` by removing the section entirely, but these two links in the library README remain broken and will 404 for anyone following them.

_{Reviews (8): Last reviewed commit: "docs(extraction): remove CLI corpus inge..." | Re-trigger Greptile}

[randerzander]

why is this version tagged ("2605")?

[kheiss-uwzoo]

The -2605 is only in the anchor target (#recommended-minimal-install-2605), not the link label. It matches the explicit { #recommended-minimal-install-2605 } on the section heading and keeps deep links from extraction docs working. The visible (26.05) on the heading itself is still there for the helm unit test that locates the minimal-install recipe. Can de-version the heading and anchor together in a follow-up eng PR.