# Search Console Audit, Pass Six: The Fixes From Pass Five Hadn't Registered Yet

> A sixth Search Console pull shows the pass-five fixes on /skills and /compare/cursor-rules-vs-skills haven't moved position yet, plus a cannibalization fix on /commands and a near-miss on the AGENTS.md vs CLAUDE.md comparison page.

Canonical: https://www.agentickit.co/blog/search-console-audit-pass-six
Last updated: 2026-10-31

**Short answer:** pulling the same Search Console export two days after [pass five](/blog/search-console-audit-pass-five) shipped was mostly a lesson in patience — `/skills` (49 impressions, position 67, 0% CTR) and `/compare/cursor-rules-vs-skills` (13 impressions, position 29.1, 0% CTR) hadn't moved, because a title-tag or FAQ-schema change doesn't re-rank a page in 48 hours. The actual work this pass was on two rows that needed a structural fix, not a waiting period: `/commands` was splitting its own "cursor commands" relevance against our own [`/blog/best-cursor-commands`](/blog/best-cursor-commands) post, and `/compare/cursor-rules-vs-agents-md-vs-claude-md` was stuck at position 57.3 for "cursor rules vs agents.md" despite an exact-match title, because the page had no internal links pointing authority at it from anywhere high-traffic.

If you've been following this series, the rule this pass adds is a boring but necessary one: before you re-pull the same rows and wonder why a fix "didn't work," check whether enough time has actually passed for a crawl-and-re-rank cycle to happen. Two days is not enough. Two to three weeks, which is what [pass one](/blog/search-console-page-audit-fixes) through [pass five](/blog/search-console-audit-pass-five) have all said at the bottom, is the actual interval. The rows worth acting on immediately are the ones with a *structural* problem — cannibalization, orphaned pages — because those don't need a re-rank to diagnose, only to fix.

## The four rows, as pulled this time

| Page | Query | Impressions | Position | CTR | Problem type |
| --- | --- | --- | --- | --- | --- |
| [/skills](/skills) | cursor skills | 49 | 67 | 0% | Repeat row from pass five — fix shipped, too early to register |
| [/compare/cursor-rules-vs-skills](/compare/cursor-rules-vs-skills) | cursor rules skills docs | 13 | 29.1 | 0% | Repeat row from pass five — fix shipped, too early to register |
| [/commands](/commands) | cursor commands | 4 | 31.8 | 0% | Cannibalization — [/blog/best-cursor-commands](/blog/best-cursor-commands) already outranks it at position 9 for the same query |
| [/compare/cursor-rules-vs-agents-md-vs-claude-md](/compare/cursor-rules-vs-agents-md-vs-claude-md) | cursor rules vs agents.md | 4 | 57.3 | 0% | Exact-match title, no internal links pointing at it — an orphan-authority problem |

## Why the first two rows are still sitting where pass five left them

This is the part worth being honest about instead of papering over: we pulled `/skills` and `/compare/cursor-rules-vs-skills` again and they're at the same impression counts and the same positions pass five recorded. That's not a sign the pass-five fix failed. Google doesn't re-crawl a page, re-evaluate a title tag, and re-rank it within 48 hours, and Search Console's own position and CTR numbers at low impression counts (13 and 49) are noisy enough that a single pull either way doesn't tell you much. We checked the code to confirm the pass-five changes are actually live — `/skills`' H1 still reads "Cursor Skills: 61 ready-to-install SKILL.md files," the FAQ block still opens with "What are Cursor skills?", and `/compare/cursor-rules-vs-skills`' comparison table is still the first thing under the lede — so there's nothing left to fix on either page from this pull. We bumped both pages' "last updated" date to register the freshness signal again and left the content alone.

The mistake this guards against is re-writing a page every time you re-pull the same Search Console row and see no movement. If the fix you shipped was the right fix, the honest move is to wait out the re-rank cycle, not to keep rewriting a title tag that hasn't had time to be re-crawled yet.

## Why `/commands` was cannibalizing its own blog post

`/commands` ranks for "cursor commands" at position 31.8. Our own [`/blog/best-cursor-commands`](/blog/best-cursor-commands) — a ranked listicle of 12 commands to copy — ranks for the same query at position 9, already on page 1. That's two AgenticKit URLs competing for one query, and the catalog page is losing. Before this pass, neither page linked to the other, so Google had no signal that one should be treated as the primary result and the other as a supporting piece.

The fix isn't to suppress the catalog page — a directory of 47 commands and a ranked listicle of 12 serve genuinely different intents (browse-everything vs. tell-me-what-to-install-first), so both deserve to exist. The fix is to make the relationship explicit instead of accidental:

1. **`/commands`** now links directly to [`/blog/best-cursor-commands`](/blog/best-cursor-commands) from its "Also read" row, labeled as the ranked pick list, right next to the two comparison pages it already linked to.
2. **`/blog/best-cursor-commands`** already linked back to `/commands` as "the commands catalog" in its closing section, so the cross-link is now two-way.
3. We left the two pages' title tags as they were — `/commands`' title ("Cursor Commands: 47 Slash Commands You Can Copy") already targets the bare "cursor commands" query directly, and the blog post's framing ("12 commands, ranked") is different enough that there's no title-level overlap to untangle. The only gap was the missing link.

Cross-linking won't single-handedly fix a cannibalization problem — Google still has to decide which URL is the better match for "cursor commands" — but an explicit link from the catalog to the post the catalog itself underperforms against gives Google's own signals something to work with instead of two unrelated-looking pages competing blind.

## Why the AGENTS.md comparison page was stuck at position 57 despite an exact-match title

`/compare/cursor-rules-vs-agents-md-vs-claude-md` has a title ("Cursor Rules vs AGENTS.md vs CLAUDE.md (2026): Which File Does Each Agent Read?") that matches the query "cursor rules vs agents.md" about as closely as a title tag can. Exact-match titles are usually the first fix in this series, which means this page had already had that fix applied before it ever showed up in Search Console — and it was still stuck at position 57.3 on only 4 impressions.

The difference from every other row in this series: nothing was wrong with the page's own content. The page already opens with a one-paragraph verdict and a compatibility matrix table inside its first couple hundred words, which is the exact structure [pass five](/blog/search-console-audit-pass-five) recommends for a relevance fix. What it didn't have was any inbound link from a high-traffic page on the site. A page can have a perfect title and a perfect lede and still sit at position 57 if nothing on the site with real authority is pointing at it — Google has comparatively little reason to trust a page's importance if the site's own internal linking doesn't reflect it.

The fix: we added a link to `/compare/cursor-rules-vs-agents-md-vs-claude-md` from both `/skills` and `/commands` — the two highest-traffic catalog pages on the site — and added `/skills` and `/commands` to the comparison page's own related-links list, so the authority flows both directions. This is the same "page ranking for two queries" logic pass five introduced, pointed at a different symptom: instead of consolidating two intents on one page, this is consolidating internal link equity onto an under-linked page that already earned its ranking on content alone.

## What we actually changed, page by page

1. **`/skills`** — No content changes; pass five's fix is confirmed live. Bumped "last updated" to 2026-10-31 and added a link to the AGENTS.md vs CLAUDE.md comparison page.
2. **`/compare/cursor-rules-vs-skills`** — No content changes; pass five's comparison table and title are confirmed live. Bumped "last updated" to 2026-10-31.
3. **`/commands`** — Added a direct link to [`/blog/best-cursor-commands`](/blog/best-cursor-commands) in the hero's "Also read" row and a link to the AGENTS.md vs CLAUDE.md comparison page; bumped "last updated" to 2026-10-31.
4. **`/compare/cursor-rules-vs-agents-md-vs-claude-md`** — Added `/skills` and `/commands` to the related-links list, and added inbound links from both of those pages; bumped the lede's internal "last updated" framing for the companion Rules-vs-Skills page in the same pass.

## The rule this pass adds to the series

**A re-pull two days after a fix ships tells you almost nothing.** Position and CTR need a crawl-and-re-rank cycle, which this series has consistently found takes two to three weeks even for small title-tag changes. If you re-pull Search Console data and a row from the prior pass hasn't moved, check the code first — confirm the fix is actually live — and if it is, leave the content alone and wait. Rewriting an already-fixed page because one early pull shows no movement just resets the freshness clock without addressing anything real.

**An exact-match title with no inbound links is still an orphan.** This series had assumed through five passes that a relevance gap was always a content or title problem. `/compare/cursor-rules-vs-agents-md-vs-claude-md` is the counterexample: perfect title, right structure, and still stuck at position 57, because nothing high-traffic on the site linked to it. Before rewriting a page's content again, check how many internal links actually point at it.

## A checklist to add to the series

1. Before re-writing a page because a re-pulled row hasn't moved, confirm the prior fix is actually live in the code, then check how long it's been — under two weeks, the data isn't ready to read yet.
2. If two of your own pages rank for the same query at different positions, link the weaker one to the stronger one explicitly, and vice versa, so Google's own signals reflect the relationship instead of guessing at it.
3. For a page with an exact-match title still stuck past position 50, check inbound internal links before touching the content again. Add links from your highest-traffic pages before assuming the copy is the problem.
4. Re-pull the same rows in 2-3 weeks, same as every prior pass — this pass is the clearest example yet of why that interval matters.

## FAQ

**Why didn't the pass-five fixes move `/skills` and `/compare/cursor-rules-vs-skills` at all?** Because only two days had passed between pass five shipping and this pull. Search Console's position and CTR data reflects what Google has already indexed and ranked; a title or FAQ change needs a re-crawl and a re-rank before it shows up, which this series has found takes two to three weeks even for small changes.

**What is cannibalization, in Search Console terms?** Two URLs on your own site ranking for the same query, splitting clicks and authority between them instead of one page capturing the query clearly. The fix is usually an explicit cross-link and, if needed, a title-level differentiation — not deleting either page, since they often serve different intents.

**Can an exact-match title still fail to rank?** Yes, if the page has no internal links from authoritative pages on the same site. Title and content answer "is this page relevant"; internal links answer "does this site think this page matters." Both need to be true.

**How is this pass different from pass five?** Pass five's two fixes were a snippet-copy (title/meta) problem and a split-ranking-across-two-queries problem. This pass's real fixes are a cross-linking cannibalization fix and an internal-link-authority fix — neither needed a content rewrite, which hadn't come up in the series before.

## Sources

- AgenticKit's own Google Search Console Performance export, pulled 2026-10-30 (first-party, not published externally)
- [We Audited Our Own Search Console Data: 4 Page Fixes That Actually Moved Position](/blog/search-console-page-audit-fixes) — the original checklist this pass extends
- [Search Console Audit, Pass Five](/blog/search-console-audit-pass-five) — the prior pull covering `/skills` and `/compare/cursor-rules-vs-skills`, confirmed still live this pass

_Written by Rakshit Yadav. Published 2026-10-31, last updated 2026-10-31._


## Related
- [rules-grader](https://www.agentickit.co/tools/rules-grader)
- [config-converter](https://www.agentickit.co/tools/config-converter)
- [seo-specialist](https://www.agentickit.co/agents/seo-specialist)
- [on-page-seo](https://www.agentickit.co/skills/on-page-seo)
- [project-conventions](https://www.agentickit.co/skills/project-conventions)
- [/seo-audit](https://www.agentickit.co/commands/seo-audit)
