gr-core

gr — Gingiris Growth main router

When to use

When the user's question falls within the "overseas growth" domain but you are unsure which sub-skill to use, start here.

Routing table

User signal	Recommended sub-skill
"Rankings dropped / can't get indexed / canonical / SEO daily report / GA4"	`gr-seo-patrol`
"Write a blog / publish an article / writing style / hreflang / Japan-Korea sync"	`gr-blog-post`
"Launch on PH / Product Hunt / hunter / maker comment"	`gr-ph-launch`
"OSS marketing / GitHub stars / open source growth / Reddit / HN"	`gr-oss-marketing`
"B2B / PLG / SLG / SaaS growth / PMF"	`gr-b2b-growth`
"ASO / App Store / UGC creators / cold start"	`gr-aso`
"User interviews / find PMF / how to ask users"	`gr-user-interview`
"Analyze competitors / competitor scan / pricing comparison / competitor blogs"	`gr-competitor`
"Split blog into Twitter / Xiaohongshu / LinkedIn / dev.to cross-post / social distribution"	`gr-social-distill`
"AI citations / GEO / llms.txt / Citable Statistics / Perplexity"	`gr-geo-cite`
"Backlinks / Wikipedia / HARO / G2 / media coverage / PR / Reddit AMA"	`gr-backlinks`

Routing decision flow

Listen to the chief complaint: What is the user's most urgent symptom? (rankings, no traffic, don't know how to write, don't know where to post)
Determine the stage: Not launched yet? (→ benchmark / PH / OSS first) Or already live? (→ SEO patrol / blog-post)
Recommend: Tell the user "I'm about to use gr-xxx, want to run it directly?"

Don't pile up all sub-skills — recommend only one at a time, and after finishing see whether cascading is needed.

Cascade recommendations (borrowed from dbskill's design)

After executing one skill, automatically recommend the next step based on the output:

gr-ph-launch 7 days after launch day → recommend gr-seo-patrol(monitor traffic)
gr-seo-patrol Cannibalization found → recommend gr-blog-post(fix canonical)
gr-blog-post After publishing → recommend gr-seo-patrol(add to daily monitoring)
gr-competitor New competitor tactic found → recommend gr-blog-post or gr-ph-launch

Reading API keys convention

All gr-* sub-skills read API keys from the same place:

Railway env vars(for production)
User's local: refer to docs/api-keys-template.md
MEMORY.md: the user's project_growth_tools.md contains the full key list

Before invoking a sub-skill, confirm whether the relevant key is available; if missing, prompt the user to configure it first — don't force it to run.

Design principles

Don't answer, dissolve the question(borrowed from dbs-diagnosis) — when the user says "my SEO dropped", first ask "which keyword, how much did it drop, when did it start" before acting
Look at the data before writing content — don't force-write without SERP / GA4 / competitor intelligence
Reuse the Iris writing style（知识库/Skill知识包/iris_writing_style.md) — all sub-skills that produce text must follow it