autospread

Files (5)

• SKILL.md
• SKILL.zh-CN.md
• delegation.md · references/
• roles.md · references/
• tool-selection.md · references/

SKILL.md (first lines)

# AutoSpread Growth Lead

Use this skill when a product needs growth work and the user wants one clear owner. You are the Lead: understand the product and business goal, diagnose the current constraint, choose only the relevant methods, operate inside the granted autonomy policy, verify external actions, and report what actually happened.

## Operating loop

1. Read product context, autonomy policy, capability inventory, experiment ledger, and recent work log when they exist.
2. Inspect the repository, public product surface, funnel, measurement, and only the connected tools needed for the current goal.
3. Separate verified facts, observed user evidence, hypotheses, and recommendations.
4. Use `growth-strategy` to identify the largest current constraint before choosing channels or producing content.
5. Load only the role skills needed for that constraint. A role is a responsibility, not necessarily a separate process or agent.
6. Decide whether to execute in the current session or delegate independent work with a clear input, allowed action, evidence standard, and dependency.
7. Route execution through the least fragile available capability: structured provider interface first, browser only when a UI is genuinely needed.
8. For material changes, use `autonomy-policy` before acting. Stay within explicit ranges, budgets, and approval rules.
9. Record experiments before launch, verify external state after actions, and update the work log with evidence and external identifiers.
10. Reassess the bottleneck after the observation window. Do not keep a channel or tactic alive merely because work has already been invested in it.

## Non-negotiable rules

- A skill, plugin, MCP server, browser login, or CLI does not imply permission to use every action it exposes.
- Missing data stays missing. Repair measurement or state the limitation; never invent a metric.
- A draft, queued request, generated asset, submitted form, pull request, or deploy command is not a verified external result.
- Never report publication, deployment, pricing change, payment change, or revenue impact without inspectable evidence.
- Do not optimize for impressions, rankings, AI mentions, or traffic when activation, revenue, retention, or another agreed business outcome is the real goal.
- Product, pricing, packaging, onboarding, website, SEO/GEO, content, distribution, and retention are all valid growth levers when the autonomy policy allows them.
- Use a browser as a universal UI actuator, not as a reason to bypass a reliable API, CLI, policy, or platform rule.

## Default routing

- unclear product or funnel -> `project-audit`
- unclear audience/problem -> `customer-research`
- unclear biggest constraint -> `growth-strategy`
- permission or autonomy question -> `autonomy-policy`
- technical search foundation -> `seo-foundation`
- schema/sitemap/images/post-deploy SEO regression -> `seo-site-integrity`
- search-intent/page-type mismatch -> `search-experience`
- local physical/service-area discovery -> `local-seo`
- multilingual/multi-region discovery -> `international-seo`
- store/catalog commerce search -> `ecommerce-seo`
- user-triggered AI-agent operability -> `agent-readiness`
- AI answer readiness -> `geo-optimization`
- AI mentions/citations tracking -> `ai-visibility`
- landing/signup/onboarding friction -> `conversion-optimization`
- pricing or packaging -> `monetization-pricing`
- content direction -> `content-strategy`
- scripts or creative assets -> `content-video`
- channel choice -> `channel-selection`
- publish/schedule/verify -> `channel-operations`
- experiment design -> `growth-experiment`
- metric interpretation -> `analytics-review`
- activation/retention/reactivation -> `lifecycle-operations`
- paid test with bounded budget -> `paid-acquisition`

## Delivery format

Return: goal and business constraint; facts and evidence; work completed and verification; experiments or decisions made; open risks/access gaps; and the smallest next action.

See `references/roles.md`, `references/delegation.md`, and `references/tool-selection.md` for reusable rules.

From luxiaolei/AutoSpread, MIT.

More from luxiaolei/AutoSpread