stale-content-detector

name	stale-content-detector
description	Use when the user says 'find old content', 'what is stale', 'what needs refreshing', or 'what should i archive'. Categorizes pages and posts as fresh, aging, stale, or archive-candidate by age, traffic signal, and internal link count, with a refresh, redirect, or archive call per item.
license	MIT
metadata	{"author":"Respira for WordPress","author_url":"https://respira.press","version":"1.0.0","mcp-server":"respira-wordpress","category":"audit"}

Stale Content Detector

Version: 1.0.0 Updated: 2026-05-24 Category: audit Status: stable Requires: Respira for WordPress plugin 7.1+ + MCP server

---

Description

Find pages and posts that haven't been updated in a long time, categorize them as fresh / aging / stale / archive-candidate, and suggest action per item: refresh, redirect, archive, or leave alone. Stale content is an SEO drag and a credibility tax — if your "2023 trends" article still ranks but nobody's touched it since 2023, you're losing trust the day a reader notices the date.

This skill is intentionally focused. It's not a content audit (use SEO & AEO Amplifier for that). It's the "what's old" question, answered fast.

---

What it categorizes

Bucket	Age since last update	Suggested action
Fresh	< 6 months	leave alone
Aging	6–12 months	review the headline + opening paragraph; if still accurate, leave; if outdated, refresh the date
Stale	12–24 months	refresh the content meaningfully (new data, new examples), update the modified date
Archive-candidate	24+ months and low traffic / orphan	archive, redirect, or 410 — depends on if anything links to it

Plus an "evergreen" override: pages tagged as evergreen (/about/, /contact/, /pricing/, /privacy/) skip the staleness check entirely.

---

When to Use

• Quarterly content review
• Migrating a site and trying to decide what to bring vs leave behind
• After a content team change — find the inherited backlog of old work
• Before publishing the next batch of new content (so the old stuff doesn't drag the rankings of the new stuff)
• After a Google update that emphasizes "freshness" signals

---

Trigger Phrases

• "find old content"
• "what's stale"
• "what needs refreshing"
• "audit my content age"
• "find content to update"
• "stale content scan"
• "what should i archive"

---

Execution Workflow

Step 1 — Confirm site

Call respira_get_active_site + respira_get_site_context.

Step 2 — Pull pages and posts

Call:

• respira_list_pages(per_page=100) for pages
• respira_list_posts(per_page=100, status=publish) for posts
• respira_list_custom_posts(post_type=<each non-default CPT>, per_page=100) for any CPT detected via respira_list_post_types

Each result includes modified and date. Sort by modified ascending — oldest first.

Step 3 — Identify evergreen pages

Pull the list of pages and check each against the evergreen list (/about, /contact, /pricing, /privacy, /terms, /legal, /imprint, /faq, /support). If the URL or slug matches, flag as evergreen and skip the staleness check.

Also check if the post type has has_archive=false and a single URL pattern matching contact/legal/utility — those are usually evergreen.

Step 4 — Compute staleness

For each non-evergreen page or post, compute days-since-modified. Bucket:

• 0–180 days → fresh
• 181–365 days → aging
• 366–730 days → stale
• 731+ days → archive-candidate (further filtering in Step 5)

Step 5 — Add traffic + orphan signals (archive-candidate filter)

For items in the archive-candidate bucket, we need to know if anyone is reading them or linking to them. Two signals:

1. Internal link count. Call respira_find_element with the page's URL as a search term. Count incoming internal links. If zero, the page is an orphan.
2. External traffic (if available — depends on whether the site has Plausible / GA4 / similar wired to Respira). If not available, skip.

Recompute:

• Archive-candidate + orphan + no traffic → archive (or 410)
• Archive-candidate + has internal links + no traffic → archive + redirect (preserve link equity)
• Archive-candidate + has traffic → refresh (downgrade from archive-candidate to stale)

Step 6 — Output the report

## Content age audit for {site_url}

Scanned {n_pages} pages + {n_posts} posts + {n_custom} custom posts ({n_total} total). Evergreen pages skipped: {n_evergreen}.

### Summary

| Bucket | Count | % |
|---|---|---|
| Fresh (<6mo) | {n_fresh} | {pct_fresh}% |
| Aging (6–12mo) | {n_aging} | {pct_aging}% |
| Stale (12–24mo) | {n_stale} | {pct_stale}% |
| Archive-candidate (24+mo) | {n_archive} | {pct_archive}% |

### Stale (12–24 months)

| Title | Last modified | Action |
|---|---|---|
| {title} | {modified} ({days_ago}d) | refresh |
| ... | | |

### Archive-candidate (24+ months)

| Title | Last modified | Internal links | Suggested action |
|---|---|---|---|
| {title} | {modified} ({days_ago}d) | {link_count} | {action} |
| ... | | | |

### What I'd refresh first (top 5 by traffic or link equity)

1. **{title}** ({days_ago}d old) — {one-line reason: high incoming link count / high traffic / topical relevance to current focus}
2. ...

Step 7 — Offer to act on a chosen item

Ask: "Want me to refresh one of these now? I can pull the existing content, suggest updates, create a duplicate with the refresh applied, and you review."

If yes, route to the SEO & AEO Amplifier skill or directly compose respira_create_page_duplicate + targeted updates.

---

Hard rules

• Never delete a page directly. Even archive-candidates get a recommendation, not a deletion. The user decides. Use respira_delete_page only if the user explicitly asks for deletion of a specific named page.
• The evergreen list is opinionated but overridable. Output the list at the top of the report so the user can see what got skipped and add/remove items.
• Don't conflate stale with bad. A 2-year-old page that still has 1,000 monthly views is not a problem. The "refresh" recommendation is based on age, not quality.
• Modified date is canonical. Don't read date (published) as a staleness signal. A 2018 post updated last month is fresh.
• Custom post types matter. Case studies, podcast episodes, properties — these often need different staleness windows. If the user runs a podcast and the latest episode is 6 months old, that's stale for podcast content even though "fresh" for blog posts. Offer to set per-CPT windows on a v1.1 of this skill.

---

Telemetry

Records: site URL hash, n_pages / n_posts / n_custom counted, bucket distribution, success/failure, total duration. No page titles, no URLs, no decisions sent.

Endpoint: POST https://www.respira.press/api/skills/track-usage