SEO Survival Kit for Claude Code

Recovery-first decision support for ecommerce/D2C SEO. Claude Code skills for Core Update recovery diagnosis, prioritized action plans, weekly monitoring, and a Change Governor / Settlement Gate that prevents over-optimizing during recovery windows. Built and validated during real ecommerce Core-Update recovery in spring 2026. Open-source, MIT, zero runtime dependencies.

v0.5.2 shipped. Public beta. Breaking changes possible before v1.0. Pin to a tag for reproducible installs. See CHANGELOG.md.

Start here

Three common entry points. Pick the one that matches your situation:

1. Traffic dropped after a Google Core Update

/seo-rescue:post-core-update-recovery example.com

Diagnosis tree + phased Authority-First recovery plan (6–12 month horizon, no 4-week magic promises).

2. Free technical health check before deciding on paid tools

/seo-rescue:seo-audit-free example.com

Free-tier check using only Google's surfaces (GSC + PSI + Lighthouse + curl). No API keys required.

3. Weekly recovery review using your Sistrix CSV exports

/seo-rescue:sistrix-monday-recovery-check current.csv previous.csv example.com

17-section structured report. CSV-first, no Sistrix API key required.

Typical sequence: diagnose → recovery plan → weekly monitoring. Every step ends with an operator-reviewable artifact. No autonomous live changes are made by any skill or command.

Positioning

Recovery-first decision intelligence for ecommerce and D2C. Most SEO tools answer what is technically wrong. This framework answers given everything I know, what should I do next, what should I leave alone right now, and what should wait. It complements technical-audit suites (claude-seo and similar); it does not replace them.

What this helps you decide

• Should this URL be protected or changed right now?
• Is this recovery signal real, or is the visibility index lagging real ranking gains?
• Are winners being offset by a few large losers, masking aggregate movement?
• Which keywords are commercially important enough to protect at all costs?
• Is now the wrong time for URL consolidation, title rewrites, or template changes?
• Should we increase paid spend, or fix margin / tracking / landing pages first?
• Which channel is profitable, which is bleeding, which to wind down?
• Which AI surfaces (AI Overview, ChatGPT, Perplexity) are currently citing the site — as a supporting signal, not a primary metric.

What this is NOT

• Not an autonomous SEO system. Every recommendation is reviewed by the operator.
• Not a full-service marketing agency. Execution belongs to the operator's team, agency, or contractors.
• Not a generic AI growth platform. The focus is recovery-first decision support for ecommerce/D2C, not breadth across all marketing surfaces.
• Not automatic ad optimization. No write actions to any ad account. No automatic budget changes.
• Not a guarantee. Every output carries calibrated confidence; the framework refuses to promise rankings, revenue, or timelines.

Safety: Change Governor and Settlement Gate

The kit is not designed to execute SEO changes autonomously. Two safety layers sit between the operator and any live modification:

Change Governor. Every session starts in audit_only mode with zero change budget. Mode escalation requires explicit operator instruction. Each proposed change is rated on risk, evidence quality, rollback readiness, and approval state — actions without a clear data basis can never be classified as green risk.

Settlement Gate. After a Major Batch of live changes, the gate hard-blocks new optimization waves until enough data has accumulated to attribute cause and effect. The most dangerous moment in recovery is not the first mistake — it is the impulse to keep optimizing immediately afterwards. The gate enforces a waiting period during which only monitoring, QA, rollback prep, and verified emergencies are allowed.

Goal. Don't change too much at once. Protect winners. Keep cause-and-effect measurable. No live SEO change without operator review.

Canonical references:

• plugins/seo-rescue/references/SEO_CHANGE_GOVERNOR.md — modes, budgets, escalation rules
• plugins/seo-rescue/references/SEO_SETTLEMENT_GATE.md — hard-block definition, exceptions, unlock criteria
• plugins/seo-rescue/references/SAFE_LIVE_CHANGE_RULES.md — pre/post-crawl rules, batch limits, user-pressure responses
• plugins/seo-rescue/docs/LIVE_CHANGE_QA.md — live-change QA workflow
• plugins/seo-rescue/commands/recovery-audit.md — change-budget audit command

Naming at a glance

The split is intentional (marketplace is the brand, plugin is the technical handle).

What's in the kit

The orchestrator (/seo-rescue:rescue) routes to the right skill or command; each can also be called directly. Detailed skill registry: ARCHITECTURE.md section 4.

Core recovery workflow

Diagnostic and analysis

Reporting and economics

AI surface (experimental visibility layer)

AI surfaces are an experimental visibility layer. Citation tracking and diagnosis only — not a guarantee of AI search traffic. Use as a supporting signal alongside classical search data, not as the sole basis for decisions.

Architecture and documentation

Installation

Pinned to a tag (reproducible, survives upstream changes):

/plugin marketplace add maxschottke-spec/seo-survival-kit#v0.5.3
/plugin install seo-rescue@seo-survival-kit
/reload-plugins

Always-latest (less safe — a maintainer-account compromise would propagate on next reload):

/plugin marketplace add maxschottke-spec/seo-survival-kit
/plugin install seo-rescue@seo-survival-kit
/reload-plugins

See SECURITY.md → How to verify before trusting.

Outside Claude Code

The three pure-Markdown framework skills (rescue, post-core-update-recovery, ai-search-rescue) are available as platform-agnostic Markdown in exports/ with install snippets for Cursor, OpenAI Custom GPT, Gemini CLI, Aider, Continue.dev, and Codex. Script-backed skills remain Claude-Code-specific; MCP wrapper planned for v1.0+.

Before you run anything

post-core-update-recovery is free. seo-outreach-report requires API credentials for Sistrix + DataForSEO + Google PSI — full walkthrough in ONBOARDING.md.

Sample output

Same layout you get for a real domain — 10 chapters, ~1 MB PDF, plain language for non-technical decision-makers:

Full sample: examples/sample-audit.pdf. The use-case pattern (sharp −60% drop in 2-3 weeks after a published Core Update) is documented in examples/screenshots/02-core-update-crash-case.png and ROADMAP-2026.md.

When to use

• Organic traffic dropped during or shortly after a published Google Core Update. Recovery framework provides a phased Authority-First plan with realistic 6-12 month horizons.
• A non-technical decision-maker (shop owner, founder, executive) needs to understand the SEO state. Outreach-report pipeline produces a decision-maker A4 PDF.
• No budget for paid SEO tools, want a free-tier health check. seo-audit-free is the entry.
• Multi-marketplace ecommerce, need to know which channel is profitable. Channel-economics-analyzer computes per-channel break-even and operating margin from your CSVs.
• Weekly PSI monitoring with regression detection. Free over PSI v5 quota.
• You want to track AI surface citation patterns (AI Overview, ChatGPT, Perplexity) as a supporting signal alongside classical rankings — not as a primary growth driver.

When not to use

• The site is brand new (< 6 months). That is an aufbau-phase, not a rescue-phase.
• You need a deep technical audit (crawlability, schema validation, CWV across hundreds of URLs). Use claude-seo instead. This complements it.
• You need 4-week magic recovery. The framework is explicit about 6-12 month horizons for Core Update damage.
• You need production-grade guarantees, SLAs, or enterprise contracts. Public beta, single-maintainer.

Use-case fit

This system is currently strongest for ecommerce/D2C recovery workflows, especially when SEO data can be evaluated against revenue, margin, inventory, returns, conversion, and paid-media context.

The framework should not be treated as a universal SEO automation system. Each website type needs its own business-signal layer.

YMYL notice

For Your-Money-Your-Life sites (medical, legal, financial, regulated), have a domain expert review the action plan before client delivery. The framework treats sites uniformly at the SEO-mechanics level; it does not validate domain-specific compliance.

Real-world basis

Skills were built from one extended Core-Update recovery case (mid-size DE ecommerce shop, March/April 2026 update window) and validated against four additional real-world domain audits in May 2026. Patterns in skill LESSONS.md files are observations from this case base, not population statistics — treat as starting hypotheses to test in your own context.

The open repo is a first-pass methodology and decision-support layer. It helps identify recovery signals, risky changes, protected URLs, winner/loser patterns, missing data, and validation steps. It is not the same as a calibrated recovery engagement that brings private business context into the decision. Deeper recovery work needs data the open repo does not see: GSC/SISTRIX history, shop revenue, margin, returns, inventory, paid-media context, implementation history, competitive context, and operator review to weigh conflicting signals.

Need help running this on your own site?

The plugin is MIT and self-serve. If you'd rather have someone with operational experience walk you through it on your specific case, the maintainer offers paid engagements:

Contact: open a Discussion or reach the maintainer via the email on the GitHub profile. The framework comes from a specific recovery case still in active recovery; engaging means getting context on which patterns apply to your situation and which do not.

Contributors

• Max Schottke — maintainer, original skills, plugin packaging
• Jeronzo — independent security review (audit 2026-05-22 surfaced the data-leak surface, prompt-injection chain, and allowed-tools hardening that landed in the v0.3.x security sprint)

Status

Single-maintainer open-source project. No SLA, no commercial support. Issue response best-effort within a few days. For production-critical workflows, pin to a specific tag and review releases before upgrading. The currently shipping version is visible in the badge near the top of this README and in CHANGELOG.md.

License, security, contributing

MIT. See LICENSE, SECURITY.md, CONTRIBUTING (open an issue first for larger changes).