GSC SEO Autopilot
Automated SEO pipeline: GSC monitoring, keyword expansion, quick wins, and blog content creation. Works with any website.
Arguments
Route based on the first argument provided by the user:
| Argument | Reference File | Description |
|----------|---------------|-------------|
| auto | references/cmd-auto.md | Full pipeline: GSC analysis + keyword expansion + blog creation |
| init | references/cmd-init.md | Interactive config setup wizard |
| weekly | references/cmd-weekly.md | GSC performance report |
| quick-wins | references/cmd-quick-wins.md | Find low-hanging SEO opportunities |
| brief <keyword> | references/cmd-brief.md | Generate content brief for a keyword |
| publish <keyword> | references/cmd-publish.md | Write a single blog post for a keyword |
Default (no argument): Run auto pipeline.
Procedure: Read the matching reference file and follow its instructions.
Config Loading
Every command MUST start by reading seo-config.yaml from this skill's directory:
~/.claude/skills/gsc-seo-autopilot/seo-config.yaml
If missing, inform user and suggest running /gsc-seo-autopilot init.
Keywords Everywhere (MCP)
Keyword data comes from the hosted Keywords Everywhere MCP server. Auth is handled by the MCP
client, so no API key is read at run time. If the mcp__keywords-everywhere__* tools are
unavailable, skip keyword expansion and fall back to WebSearch -- never fail the run.
Tools
| Tool | Required args | Returns | Credits |
|------|--------------|---------|---------|
| get_related_keywords | keyword, num | data: array of keyword strings | 2 per keyword returned |
| get_pasf_keywords | keyword, num | data: array of "People Also Search For" strings | 2 per keyword returned |
| get_keyword_data | kw, country, currency, dataSource | data[]: {keyword, vol, cpc:{currency,value}, competition, trend[]} | 1 per keyword |
| get_domain_keywords | domain, country, num | data[]: {keyword, estimated_monthly_traffic, serp_position} | 2 per keyword |
| get_credit_balance | -- | array with one integer | free |
Every argument is required -- none have defaults. Note that get_related_keywords and
get_pasf_keywords take only keyword and num; country, currency, and dataSource
do not apply to them and are used solely by get_keyword_data.
Limits and cost
get_keyword_dataaccepts a maximum of 100 keywords per call. Chunk longer lists.numaccepts up to 10,000, but expansion costs 2 credits per keyword returned. Passke_related_limitfrom config -- never a large literal.- Check
get_credit_balancebefore a run that expands many seeds.
Map config values to arguments: ke_country -> country, ke_currency -> currency,
ke_data_source -> dataSource, ke_related_limit -> num.
Blog Writing
Follow references/blog-writing-guide.md. Do NOT delegate to external skills.
Setup
- GSC MCP server (see README)
- Keywords Everywhere MCP server (optional, see README)
- seo-config.yaml in skill directory (run
/gsc-seo-autopilot initor copyseo-config.example.yaml)
微信扫一扫