You can run the scripts in CI as a reporting step, but the current compare script exits nonzero only on an execution error. Critical findings are printed to stderr and still return success. A tested external severity gate is required before the command can fail a build on critical drift. Baseline IDs, timestamps, and local cache persistence also need explicit CI handling.

// SKILL

SEO DRIFT MONITORING
GIT FOR ON-PAGE SEO

SEO drift is the gradual loss of search-critical page signals after site changes. A deploy can remove a meta description, rename an h1, strip schema, or flip a canonical. Claude SEO records a baseline, compares any future state, and reports what changed and how severe the regression is.

By Daniel Agrici · Command surface verified: 2026-08-27

CONTRIBUTED BY DAN COLTA · PRO HUB CHALLENGE · original repo

$

/seo drift baseline https://your-site.com

REQUIRES CLAUDE SEO INSTALLED IN CLAUDE CODE

// RUN PREVIEW

What will happen if you start this skill?

The workflow stores a defined local baseline and reports changes; severity does not make the current utility a CI gate.

OUTCOME

A baseline, comparison, or history report across the fields and 17 rules implemented by the current utilities.

REQUIRED INPUT

Use /seo drift baseline <url>, /seo drift compare <url>, or /seo drift history <url>.

READINESS

Needs a reachable public page and persistent local cache. Optional PageSpeed evidence needs compatible credentials and availability.

LOCAL OUTPUT

JSON results plus history in ~/.cache/claude-seo/drift/baselines.db. Select an older record with --baseline-id.

SIDE EFFECTS

Fetches the page and writes baseline or comparison records to the user cache. It does not edit the monitored site.

COST + APPROVAL

Claude Code and optional PageSpeed access can cost separately. Define and test any external severity gate before it blocks deployment.

Listen to the SEO drift summary

A 62-second spoken orientation to baseline capture, local storage, comparison limits, CI gating, and operating requirements.

1:02 AUDIO

AI-generated narration using Daniel's authorized voice at 1.1x. The user-supplied music bed is mixed at 0.02 linear gain with a 2-second fade in and a 3-second fade out. An integrated-loudness comparison measures the speech stem 23.69 LU above the music stem. The 20 LU internal mix target is an engineering proxy, not a WCAG conformance measurement. Transcript alignment is 0.9401 against the intended script. This 62.428-second candidate passed technical checks. Human listening review is pending, so the audio is not approved for release.

// HOW IT WORKS

How does SEO drift monitoring work?

SEO drift is an unintended change to tracked page elements between captures. A title can disappear after a CMS change, a canonical can point to a different URL, a noindex directive can appear, or a JSON-LD block can be removed. The comparison makes those changes reviewable; it does not prove that a finding caused ranking loss.

The drift skill captures a defined set of fields in local SQLite, assigns the baseline a numeric ID and timestamp, and compares a later fetch with the latest or a selected baseline. It reports critical, warning, and info findings. The current script does not turn critical findings into a failing process exit.

Primary references: Google's Search Analytics API documentation explains the performance data used to investigate traffic changes, and the URL Inspection API reference documents programmatic index-status checks.

The Six Pillars

Drift monitoring is built on six load-bearing capabilities. Each is described below.

01
BASELINE CAPTURE
One command captures the supported fields in SQLite and returns a numeric baseline ID. Use the most recent baseline or select an ID at runtime.

02

17 COMPARISON RULES

Rules cover title, meta description, canonical, noindex, h1, h2, schema, Open Graph, status, optional CWV, and content hashes.

03

SEVERITY TIERS

Critical (fix now): canonical flip, noindex added, or h1 removed. Warning (review): title or meta description changed. Info (track): h2 structure or full content hash changed.

04

SQLITE HISTORY

Baselines and comparison summaries persist in the user's local cache. History lists stored records for the normalized URL.

05

CI INPUT

The scripts can produce CI artifacts, but severity does not control the current exit code. Add and test a separate gate before blocking a build.

06

DEPLOYMENT GUARD

Pair with claude-seo audit for full-stack confidence: drift catches the small stuff, audit catches the systemic stuff.

// USAGE

How do you run SEO drift checks?

Commands

Command	What it does
`/seo drift baseline <url>`	Capture the supported page fields and store a timestamped baseline with a numeric ID
`/seo drift compare <url>`	Compare the current fetch with the most recent stored baseline
`claude-seo run drift_compare.py <url> --baseline-id N`	Use a specific numeric baseline ID through the script runtime
`/seo drift history <url>`	List stored baselines and comparison summaries for the URL

Severity Tiers

Every triggered rule has one of three severity levels. Severity drives the report summary only; it does not currently change the process exit code.

Tier	Meaning	Examples
CRITICAL	High-priority review in the skill rubric	Canonical changed or removed, noindex added, h1 or title removed, schema removed, status changed to 4xx/5xx
WARNING	Potential regression to investigate	Title or meta description changed, CWV regressed, performance score dropped, OG tags removed, schema modified
INFO	Awareness item that may be intentional	Schema added, h2 structure changed, or the full HTML content hash changed

CI Boundary

The scripts can run in CI and emit JSON, but drift_compare.py exits nonzero only when execution returns an error. Critical, warning, and info findings are printed while the process still succeeds. To block a build, add a separate tested step that parses the JSON summary and applies your approved severity policy.

claude-seo run drift_baseline.py https://example.com
claude-seo run drift_compare.py https://example.com --baseline-id 5

The SQLite database lives at ~/.cache/claude-seo/drift/baselines.db, not in the project root. A CI job must provide persistent storage or an explicit baseline handoff if history needs to survive between runs. Baselines include timestamps and numeric IDs, so repeated captures are not byte-identical.

What Gets Captured

Each baseline records the SEO-critical fields below. The 17 comparison rules operate on subsets of these fields.

Title tag and meta description
Canonical URL and meta robots
Heading hierarchy: h1, h2, h3 arrays
JSON-LD schema blocks (parsed and hashed)
Open Graph tags
Core Web Vitals via PageSpeed Insights (optional, skip with --skip-cwv)
HTTP status code
SHA-256 hashes of full HTML body and schema content for fast change detection

The current baseline does not store hreflang, Twitter Card tags, image alt text, internal anchor distribution, or h4-h6 headings.

// FAQ

What should you know about SEO drift?

What is SEO drift?

SEO drift is an unintended change to tracked page elements between captures. The current implementation can report changes such as a removed title, changed canonical, added noindex directive, removed schema, changed headings, or an error status. A finding is evidence for review, not proof of ranking loss.

What elements does it baseline?

How is history stored?

The local path ~/.cache/claude-seo/drift/baselines.db stores baselines and comparisons. Baselines receive numeric IDs and timestamps; there is no label field. Pass --baseline-id N to use an older record; otherwise compare uses the most recent baseline. History lists stored baselines and comparison summaries.

Can I run it in CI?

Who contributed this skill?

// RELATED SKILLS

EXPLORE MORE

VIEW ALL 26 SKILLS →

// FROM THE BLOG

SEO DRIFT MONITORING
GIT FOR ON-PAGE SEO

What will happen if you start this skill?

How does SEO drift monitoring work?

The Six Pillars

How do you run SEO drift checks?

Commands

Severity Tiers

CI Boundary

What Gets Captured

What should you know about SEO drift?

EXPLORE MORE

RELATED GUIDE

CATCH SEO REGRESSIONS
BEFORE THEY SPREAD.

SEO DRIFT MONITORINGGIT FOR ON-PAGE SEO

What will happen if you start this skill?

How does SEO drift monitoring work?

The Six Pillars

How do you run SEO drift checks?

Commands

Severity Tiers

CI Boundary

What Gets Captured

What should you know about SEO drift?

EXPLORE MORE

TECHNICAL SEO

FULL SITE AUDIT

SCHEMA MARKUP

RELATED GUIDE

SEO DRIFT MONITORING DEEP DIVE

CATCH SEO REGRESSIONSBEFORE THEY SPREAD.

SEO DRIFT MONITORING
GIT FOR ON-PAGE SEO

CATCH SEO REGRESSIONS
BEFORE THEY SPREAD.