Skip to content
Nextrank
Start free

Reference

Agent Skills

Nine pre-built Claude Agent Skills that drive the Nextrank MCP server — install one, and an agent runs the whole workflow (orient, budget, research, track, audit, review) on its own.

An Agent Skill is a packaged workflow: instructions an agent follows, plus the specific MCP tools it calls, in the right order, with the right guardrails around credits and rate limits. Nextrank ships nine of them, covering orientation, budget checks, research, tracking, technical audits, and real search data. Install the ones you want — they are not bundled into a plugin.

Orientation

seo-coach

Looks at a project — saved keywords, a tracker, a Search Console connection — and recommends exactly one next skill, in priority order: keywords before tracking, tracking before review, review before a competitor check. Spends no credits; it only reads.

Say things like "help me with SEO for acme.com" or "I just signed up, what should I do first?"

credit-budget-check

A pre-flight check for a batch of paid calls, not a substitute for the per-call cost line every paid skill already gives you. Reads the org's credit balance, estimates the planned calls with nextrank:estimate_research and nextrank:estimate_rank_run, sums what can be estimated precisely, and warns before anything runs if the batch would exceed what's available. Spends no credits itself.

Say things like "do we have budget for keyword research on all five projects" or "check we can afford this before you start".

Research

keyword-research-plan

Turns a domain or topic into a scored, saved keyword shortlist while spending the fewest credits possible. Estimates cost with nextrank:estimate_research before ever calling the paid nextrank:research_keywords, clusters the results into 3–6 themes, and saves 10–20 keywords with a single prioritised recommendation on which to target first.

Say things like "plan keyword research for acme.com" or "what should this blog rank for".

keyword-metrics-refresh

Maintenance, not discovery: pulls fresh volume, difficulty, CPC and intent for keywords already saved to a project, batched 100 at a time, with a 24-hour cache so re-running it costs nothing extra. Use it before a review or a client call when a saved list is a few weeks old and might have moved.

Say things like "refresh our keyword metrics" or "is this volume data still right?"

Tracking & review

rank-tracker-setup

Turns a saved keyword list into a running tracker with one check already executed, so weekly-rank-review has data on day one. Spends no credits — rank tracking draws from the plan's keyword-slot pool instead.

Say things like "set up rank tracking for acme.com" or "track these 10 keywords weekly".

weekly-rank-review

Reads a tracker's movement over the last period and writes a plain-language, forwarding-ready summary — wins, watch-list, and exactly one prioritised Recommendation. Needs a tracker with at least one completed run; point the agent at rank-tracker-setup first if there isn't one yet.

Say things like "how did our rankings move this week" or "put together a ranking update for the client".

Technical audit

site-audit-triage

Starts (or reads an existing) crawl of a project's domain and turns the raw issue list into a short, prioritised, plain-language fix list led by the highest-impact CRITICAL issues — instead of dumping every one of the ~35 issue types with equal weight. The first 50 pages of every audit are free on every plan; beyond that it costs 1 credit per 2 pages, reserved as an estimate at start and settled once the crawl finishes. Starting a crawl returns immediately — the agent polls status rather than waiting.

Say things like "audit acme.com" or "what's broken on our site".

Real search data

search-console-quick-wins

Pulls real Google Search Console performance — clicks, impressions, CTR, average position, with 28-day deltas — and surfaces the queries sitting at position 8–20 with real impressions: the cheapest rank gains available, because the page is already close. Uses no credits; needs Search Console connected on the project first (that one step is web-only — an agent can't complete the browser consent).

Say things like "find quick wins for acme.com" or "what's almost on page one?"

Competitive insight

domain-snapshot

A fast competitive read on one domain, or up to three domains compared against it: Standing Score, organic traffic, ranked keywords, referring domains, top pages, and — with competitors — a Verdict on whether backlinks or content is the bigger gap. Charges per domain looked up, cached for 24 hours.

Say things like "how's acme.com doing" or "compare us to competitor.com".

All nine, at a glance

SkillCategoryCredits
seo-coachOrientationFree
credit-budget-checkOrientationFree
keyword-research-planResearchEstimated first, then charged
keyword-metrics-refreshResearchCharged, batched + cached
rank-tracker-setupTracking & reviewFree — uses keyword slots
weekly-rank-reviewTracking & reviewFree
site-audit-triageTechnical auditFirst 50 pages free, then charged
search-console-quick-winsReal search dataFree
domain-snapshotCompetitive insightCharged per domain

Install

Claude Code — copy a skill folder into ~/.claude/skills/ (personal) or .claude/skills/ (project):

npx skills add github.com/VladanIlicJS/agentseoplatform --skill seo-coach
npx skills add github.com/VladanIlicJS/agentseoplatform --skill credit-budget-check
npx skills add github.com/VladanIlicJS/agentseoplatform --skill keyword-research-plan
npx skills add github.com/VladanIlicJS/agentseoplatform --skill keyword-metrics-refresh
npx skills add github.com/VladanIlicJS/agentseoplatform --skill rank-tracker-setup
npx skills add github.com/VladanIlicJS/agentseoplatform --skill weekly-rank-review
npx skills add github.com/VladanIlicJS/agentseoplatform --skill site-audit-triage
npx skills add github.com/VladanIlicJS/agentseoplatform --skill search-console-quick-wins
npx skills add github.com/VladanIlicJS/agentseoplatform --skill domain-snapshot

Or clone the repo and copy folders directly:

git clone https://github.com/VladanIlicJS/agentseoplatform
cp -r agentseoplatform/skills/seo-coach ~/.claude/skills/

claude.ai — zip a skill folder (SKILL.md at its root) and upload it under Settings → Features → Skills. Custom skills don't sync across surfaces, so repeat this on each surface you use.

Claude API — upload the same zip through the /v1/skills endpoints (requires the code execution tool enabled on the key).

What next