| name | livepolls-subscription-benefit-research |
| description | Re-run the final LivePolls subscription benefit-copy audit and generate the actionable HTML used for AI-assisted pricing-page fixes. Use when the user asks to recheck LivePolls pricing pages after modifications, regenerate livepolls_subscription_benefit_research.html, audit weak subscription benefits, compare product pricing benefits against competitor pricing pages, or produce a direct product-by-product HTML fix guide focused on paid-plan benefits rather than localization, FAQ, unit, browser-rendering, or /api/order/plans issues. |
LivePolls Subscription Benefit Research
Purpose
Generate the final actionable HTML:
livepolls_subscription_benefit_research.html
This skill is for the user-facing "权益文案 / 付费卡转化点" audit. It intentionally avoids the earlier broad localization report unless the user asks for that separately.
Scope
In scope:
- Read the existing
livepolls_pricing_audit.html produced by the LivePolls page-data collector.
- Extract
audit-data from that HTML.
- Detect products whose paid-plan benefit copy is weak, generic, or confusing.
- Group products by type.
- Add competitor pricing-page evidence.
- Output a compact HTML with clickable product links, concrete issues, and direct fix suggestions.
Out of scope for this skill:
- Real browser language-switch verification.
- Multi-language JSON matrix review.
- FAQ translation quality review.
unit/plural_unit localization audit.
- Paddle/payment/login/browser-rendering behavior.
- Re-querying
/api/order/plans directly from guessed app-specific paths.
If the user asks for those, use the localization/page-data audit workflow instead.
Required Input
Default input in the current working directory:
livepolls_pricing_audit.html
It must contain:
<script id="audit-data" type="application/json">...</script>
If the file is missing, first locate or regenerate the page-data collector output. Do not invent product rows.
Script
Bundled script:
scripts/build_subscription_benefit_research.mjs
Default behavior:
node scripts/build_subscription_benefit_research.mjs
When running outside the skill folder, call it by absolute path from the target workspace:
node /Users/Dario/.codex/skills/livepolls-subscription-benefit-research/scripts/build_subscription_benefit_research.mjs
Optional path overrides:
LIVEPOLLS_AUDIT_HTML=/path/to/livepolls_pricing_audit.html \
LIVEPOLLS_OUTPUT_HTML=/path/to/livepolls_subscription_benefit_research.html \
node /Users/Dario/.codex/skills/livepolls-subscription-benefit-research/scripts/build_subscription_benefit_research.mjs
The script fetches competitor pricing-page text through curl and r.jina.ai reader URLs. If network access is blocked, request approval and rerun. Fallback evidence exists for some competitors, but live fetches are preferred.
Special pricing URL handling:
convert uses https://convert.livepolls.app/pricing.
recorder uses https://recorder.livepolls.app/pricing.
translate uses https://translate.livepolls.app/pricing.
https://www.imgkits.com/pricing and https://www.livepolls.app/pricing are standalone/root pricing pages. Include them as special-entry reminders, but do not count them as extra app-key rows in the 255-product audit.
- Do not use the root
https://www.livepolls.app/api/order/plans response as the套餐 source for arbitrary product slugs; it returns the root Live Polls Maker plan, not each product's plan.
Workflow
-
Confirm the target workspace and the intended output path.
-
Verify input exists:
ls -lh livepolls_pricing_audit.html
-
Run syntax checks before changing or running the script:
node --check /Users/Dario/.codex/skills/livepolls-subscription-benefit-research/scripts/build_subscription_benefit_research.mjs
-
Generate the final HTML from the workspace containing livepolls_pricing_audit.html.
-
Verify the command output reports product count, benefit issue count, category count, competitor pages, and successful competitor fetches.
-
Spot-check the HTML:
rg -n "Livepolls 订阅页权益文案修复清单|统一修改原则|网址|相应问题|修复方案" livepolls_subscription_benefit_research.html
-
Open or hand off only the final HTML unless the user asks for intermediate files.
Expected Output Shape
The HTML should be self-contained and designed for AI-assisted modification:
- Header with total product count, issue count, category count, and data source.
- Search box.
- Category navigation.
- Product-type sections.
- Each row has:
- Clickable pricing URL.
- Specific problem in current benefit copy.
- Product-specific repair direction.
- Suggested fields to update, usually
monthPlans.features, yearPlans.features, and lifetimePlans.features if present.
- Competitor evidence cards for the category.
- A
特殊 pricing 入口 section listing the known special URLs and whether they belong to app-key rows or standalone/root pricing pages.
Detection Rules
Flag benefits when they contain or resemble:
Unlimited use as the main paid benefit.
- Discount/price/timer text such as
Save up to 50%, Pro Monthly, Flash Sale.
- Free-plan limitation text inside paid benefits, such as
Can only generate 1 image.
- Too few feature bullets.
- Missing concrete signals such as quota, export quality, watermark, commercial use, supported formats, speed, support, templates, models, history, pages, files, credits, minutes, or collaboration.
Do not claim a capability as fact unless it is already in source data. Use [N] placeholders and "if supported" wording when the real entitlement must be confirmed by the product owner.
Final Response
Reply in Chinese and keep it short:
- Link the generated HTML with a local file link.
- State the counts from the script output.
- Mention if competitor fetches failed or network fallback evidence was used.
- State that this skill covers subscription benefit copy only, not language switching or payment rendering.