Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add maxschottke-spec/seo-survival-kit --skill gsc-deep-divegit clone --depth 1 https://github.com/maxschottke-spec/seo-survival-kitWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/maxschottke-spec/seo-survival-kit/gsc-deep-dive)<a href="https://agentmods.dev/skills/maxschottke-spec/seo-survival-kit/gsc-deep-dive"><img src="https://agentmods.dev/badge/skills/maxschottke-spec/seo-survival-kit/gsc-deep-dive/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/maxschottke-spec/seo-survival-kit/gsc-deep-dive"><img src="https://agentmods.dev/badge/skills/maxschottke-spec/seo-survival-kit/gsc-deep-dive.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00143 | $0.03761 |
| Opus 5 | $0.00072 | $0.01880 |
| Sonnet 5 | $0.00029 | $0.00752 |
| Haiku 4.5 | $0.00014 | $0.00376 |
Grade B, and why
gsc-deep-dive scanned grade B with 3 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Instruction-override phrasingmediumPrompt injection
Text telling the model to disregard its earlier instructions or safety rules is the shape of a prompt injection, whoever wrote it.
- GSC API responses pass through `sanitize()` for the `query` strings — those are user-provided search queries from Google's index and are *attacker-controllable* in the sense that someone can search for `Ignore previous Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.
Asks for rootlowPrivilege escalation
A mod that escalates privileges can change anything on the machine, not only the project.
chmod 600 ~/.config/seo-rescue/gsc-service-account.json Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
allowed-tools: [Read, Write, Bash(node:*), Bash(curl:*)] How it starts
The opening of the file, as written. The whole thing — 234 lines — stays where its author put it; the contents beside it link to each section on GitHub.
GSC Deep Dive
Overview
Google Search Console is the canonical source for what users actually search and click on your site, plus Google's own assessment of indexability, manual actions, security, and Core Web Vitals field data. Most workflows force the user to manually open GSC, click through 5-7 reports, screenshot or CSV-export each, and paste into the Claude context.
This skill removes that friction: one config (service-account JSON or OAuth token), one command, full JSON dump. From there, [[post-core-update-recovery]], [[seo-outreach-report]], and [[ai-search-rescue]] consume the data directly.
When to use
- A site running active recovery — pull GSC weekly to track query / page / impression drift in one report
- An audit handoff — gather every GSC signal in one JSON before invoking [[seo-outreach-report]]
- Verifying a recommendation actually landed — re-pull after 2-4 weeks and diff against the previous snapshot
- A YMYL / B2B site — manual actions + security issues + indexation status are the YMYL-relevant signals
- An agency onboarding a client — first-call data gathering in <5 minutes
Don't use for:
- Real-time monitoring (GSC data lags ~2 days for fresh queries, ~7 days for impressions)
- Competitor data (GSC only shows your own property)
- Backlink analysis (GSC has a backlink view but [[seo-outreach-report]] / [[competitor-deep-audit]] pull DataForSEO's more complete dataset)
What gets pulled
Per run, the script writes a single JSON file containing:
| Field | Source | What it tells you |
|---|---|---|
site |
config | The verified GSC property |
period |
start_date / end_date |
The window queried (default last 90 days) |
top_queries |
searchanalytics.query (rows=500) | Top 500 queries by clicks: keyword, clicks, impressions, CTR, position |
top_pages |
searchanalytics.page (rows=500) | Top 500 pages by clicks: URL, clicks, impressions, CTR, position |
query_page_pairs |
searchanalytics.query+page (rows=1000) | What page ranks for what query (most useful for cannibalisation detection) |
coverage |
urlInspection if site-level | Indexed vs not-indexed counts |
last_crawl_dates |
urlInspection per top-10 URL | When Google last crawled each top page |
manual_actions |
searchconsole.searchanalytics + sites.list | Any pending manual actions on the property |
crawl_errors_summary |
searchanalytics aggregates | Page-level errors (4xx, 5xx, soft-404) the last 30 days |
crux_field_data |
from PSI v5 (cross-reference) | LCP / CLS / INP per top URL — field data, not lab |
search_appearance |
searchanalytics with searchAppearance dimension | SERP feature breakdown (AMP, video, FAQ, How-to, Sitelinks, AI Overview where exposed) |
query_weekly_series |
searchanalytics date+query, ISO-week-bucketed, top 200 by clicks | Per-query weekly click series over weekly_series_days (default 480) — input for quiet-death detection |
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 234 lines · 143 tokens per session scan B 475be7020a41
gsc-deep-dive is a skill published in the GitHub repository maxschottke-spec/seo-survival-kit (9 stars, last pushed 1mo ago), licensed MIT. It adds 143 tokens to every session and 3,761 once invoked, about $0.0007 per session on Opus 5. A static security scan graded it B with 3 findings (instruction-override phrasing, asks for root, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
content-seo
When the user wants to brief, write, review, optimize, or publish public SEO content. Also use for article outlines, blog posts, landing-page copy, refreshes, and ranking-oriented editorial drafts.
brain-keeper
When the user wants to ingest sources, change brain pages, register decisions, catalog content publications, or lint brain pages for provenance and link integrity.
backlink-analysis
When the user wants backlink, referring-domain, anchor, link-quality, link-gap, link-intersect, anchor-diff, link-velocity, brand-mention, or competitor link-profile analysis for one target domain or URL.
competitive-analysis
When the user wants to compare domains or URLs against competitors across SEO footprint, share of voice, keyword/content gap, head-to-head pages, off-page link gap (via backlink-analysis), and brand positioning. Orchestrates evidence from serp-extract, keyword-research, backlink-analysis, and topic-cluster without…
start
Friendly first-run alias that routes new Agentic SEO users to the canonical agentic-seo and project-init workflow without duplicating process rules.
autoresearch
When the user wants a rigorous iteration loop for an artifact, prompt, briefing, content structure, or Agentic SEO skill. Also use for Karpathy-style experiment runs that need baseline scoring, explicit metrics, stop rules, and keep/reject decisions.