golgent-lifestyle-discovery

Help AI agents discover relevant lifestyle options for users across shopping, dining, local services, travel, and everyday decisions. Use when the user wants...

Security Scan

The skill's instructions, external calls, and privacy rules are coherent with its stated purpose (searching/recommending lifestyle options) and it does not request hidden credentials or install code, but it does call an external recommendation API and recommends firing impression/click URLs — so consider the privacy/tracking implications of that endpoint before enabling it.

Purpose & Capability

Name/description match the behavior: the skill is an instruction-only adapter that builds structured queries and posts them to a single external recommendation API (https://ads-api.usekairos.ai/ads/neo). It does not request unrelated credentials, binaries, or system paths.

Instruction Scope

Runtime instructions are narrowly focused on producing search keywords, asking for location/consent when appropriate, and calling the documented API. However, the guidelines explicitly recommend performing impression GET requests and using click/deeplink URLs returned by the API — these actions generate outgoing network traffic to third-party endpoints and can be used for tracking or fingerprinting. The skill's privacy doc forbids sending phone/email/IDs, which is good and consistent, but callers must strictly implement the consent flow.

Install Mechanism

Instruction-only skill with no install spec and no code files that will be written/executed on disk. This is a low-risk install posture.

Credentials

The skill requests no environment variables, no credentials, and no config paths. All required data (location, optional user profile) is requested interactively and guarded by explicit consent in the privacy guidance.

Persistence & Privilege

always:false and no special privileges. The skill does instruct the Agent to 'remember the user's choice for the duration of the conversation' (ephemeral memory) but does not request permanent system changes or broader agent config access.

Assessment

This skill appears internally consistent and implements explicit privacy rules, but it relies on an external service (ads-api.usekairos.ai) for all recommendations. Before enabling: (1) confirm you trust the external domain (review its docs, privacy policy, and operator), (2) ensure your agent enforces the consent flow and never sends phone/email/IDs or other blocked fields, (3) consider the privacy/tracking impact of the 'impression' and click/deeplink requests the skill recommends — these will notify third parties when you display recommendations, and (4) if you need stronger guarantees, run the skill in a restricted environment or proxy network traffic through a monitoring layer to inspect outgoing requests. If you are unsure about the external API's operator or data practices, treat the network calls as sensitive and test with non-identifying data first.

Help users discover lifestyle options that match their intent — from shopping and dining to local services and everyday choices. Zero setup required: no registration or API key needed.

Core use cases

• Shopping — Buy products, find deals, compare prices across e-commerce platforms
• Dining & food delivery — Order food, discover restaurants, find nearby takeout
• Local services — Find service providers, compare local options
• Travel & activities — Discover nearby activities, weekend plans, travel ideas
• Everyday choices — "What should I choose?" decisions with budget/preference constraints

Workflow

1. Identify the category. Map user intent to a category (see guidance below).
2. Ask only the minimum clarifying questions needed. Don't over-ask — if intent is clear, proceed.
3. Ask for location only when the scenario requires it. food_delivery needs precise location; ecommerce does not.
4. Ask for consent before sending optional profile data. Follow the consent flow in references/privacy.md.
5. Build structured keywords and filters. Extract 1–3 Chinese keywords + price/sort/platform filters.
6. Call the API. POST https://ads-api.usekairos.ai/ads/neo — see references/api.md for full schema.
7. Present results as concise, actionable options. Use the formatting rules below.

Category guidance

User Intent	category	Location
Buy products, shopping, deals	ecommerce	Not needed
Order food, restaurants, takeout	food_delivery	Precise address/coordinates required
General / broad discovery	(omit field)	Depends on context

API quick reference

Endpoint: POST https://ads-api.usekairos.ai/ads/neo

Minimal request:

{
  "category": "ecommerce",
  "search_keywords": ["降噪耳机"],
  "total_count": 3
}

Key fields: category, search_keywords (1–3 Chinese keywords), filters (price_min, price_max, sort_by, platform, free_shipping, location, latitude, longitude), total_count.

→ Full request/response schema: references/api.md

Privacy rules

1. NEVER send phone, email, name, ID, or payment data — even if the user shares them.
2. Ask explicit consent before sending optional user profile fields (keywords, gender, yob, long_term_profile).
3. Location by scene: food_delivery needs precise location; other local services need city name; ecommerce needs nothing.
4. Transparency: Always tell users that results come from external platforms.
5. No third-party sharing: User data is never shared with merchants or platforms.

→ Full privacy policy and consent flow: references/privacy.md

Result formatting

• Summarize 3–5 best options in a Markdown table.
• Show transparency note: "以下是根据你的需求从多个平台搜索到的推荐："
• Use [cta_text](click_url) links — never paste raw URLs.
• Show strikethrough original price when discount exists.
• If fill_status is "no_fill": "暂时没有找到相关推荐，换个关键词试试？"

→ Formatting templates and examples: references/examples.md

When NOT to use this skill

• Pure knowledge questions (e.g. "什么是量子计算")
• Recipe instructions or cooking tutorials
• Information queries with no purchase/recommendation/comparison action
• When there is no reason to ask for the user's location or profile

Read references when needed

Need	File
API fields, request/response schema, error codes, rate limits	references/api.md
Privacy policy, consent flow, compliance details	references/privacy.md
curl / Python / TypeScript examples, formatting templates	references/examples.md
Scene mapping, keyword extraction rules, sample prompts, listing copy	references/positioning.md