agentsclimarketplace
Sign in

Capture competitor promotions

Skill ljh1097912305-wq/competitor-promo-agent-kit/skills/capture-competitor-promotions

Capture current promotional actions from competitor official websites, collect evidence screenshots, and create an Excel comparison table with screenshots embedded in each brand row. Use for competitor promotion audits, official-site campaign monitoring, current sale checks, 竞品官网促销截图、官网活动对比、Prime Day 或节日大促监测, and evidence-backed promotion spreadsheets.From its SKILL.md

Install
npx -y skills add ljh1097912305-wq/competitor-promo-agent-kit --skill capture-competitor-promotions

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
  • runs commandsInstructs the agent to run 2 commands, including `promo-audit capture --input capture.json --output ./data/capture` and 1 more.

SKILL.md

3.8 KB, 688 tokens by cl100k_base, as published. Nobody here has run it

Capture Competitor Promotions

Create a current, evidence-backed comparison of promotions shown on official brand websites.

Evidence rules

  • Use official brand-owned websites as evidence. Do not use marketplaces, search snippets, affiliate pages, or news articles as proof of the current website promotion.
  • Describe the promotion actually visible on the captured page. Do not force a Prime Day, holiday, or campaign association unless the official page explicitly makes it.
  • Capture one screenshot when it proves the action. Use a second or third only when a separate page or viewport is necessary. Never exceed three screenshots per brand.
  • Record exact visible claims. Do not infer discount amounts, dates, gifts, promo codes, or shipping terms.
  • If the captured official pages show no explicit promotion, write No explicit promotion found on captured official pages and preserve the evidence.
  • Keep each evidence URL and capture timestamp with the brand record.

Workflow

  1. Resolve and verify each official website URL.
  2. Capture the homepage first with capture_promotion_pages.
  3. Inspect the returned screenshot, visible text, final URL, and promotion-link candidates.
  4. If the homepage is sufficient, stop capturing that brand. If needed, capture the most relevant official promotion page. Keep the total at three or fewer.
  5. Convert only verified claims into structured promotion records.
  6. Call build_promotion_report so the screenshots are embedded in the corresponding brand row.
  7. Return the workbook and the capture manifest. Mention capture errors or unavailable sites explicitly.

Tool use

capture_promotion_pages

Pass exact official URLs and at most three page specifications per brand. Prefer a normal viewport screenshot. Use selector only when a stable element clearly contains the complete promotion.

The tool returns:

  • screenshot artifacts;
  • requested and final URLs;
  • page title and capture time;
  • normalized visible page text;
  • promotion-link candidates;
  • a JSON capture manifest.

Treat promotion-link candidates as navigation hints, not as verified promotion claims.

build_promotion_report

Create one record per brand with:

  • brand
  • officialUrl
  • promotionName
  • mechanisms
  • discounts
  • gifts
  • priceExamples
  • promotionPeriod
  • promoCode
  • shipping
  • evidenceScreenshots
  • evidencePages
  • capturedAt
  • notes

Use screenshot paths returned by the capture tool. The workbook must keep screenshots in the matching brand row.

read_promotion_artifact

Use this when a remote MCP client needs the generated JSON, screenshot, or workbook returned as an embedded resource instead of a server-local path.

CLI fallback

When MCP is unavailable but shell execution is available:

promo-audit capture --input capture.json --output ./data/capture
promo-audit report --input report.json --output ./data/comparison.xlsx

Read the manifest contract before constructing report records manually.

Quality check

Before delivery, verify:

  • every brand has an official URL;
  • every claim is supported by captured page content;
  • each brand has one screenshot when possible and no more than three;
  • screenshots appear in the correct Excel row;
  • evidence URLs and timestamps are retained;
  • missing or blocked evidence is labeled rather than guessed.

What ships with it: 2 files

2.8 KB alongside SKILL.md

agents/

references/

Gives 0 of the 12 instructions most marketing audience skills give in 688 tokens

Counted across 690 of the 894 authors here whose files we hold, read 2026-08-07

  • Apply Poppins font to headingsin 41 of 690, across 6 files
  • Apply Lora font to body textin 41 of 690, across 6 files
  • Use Arial fallback for headingsin 39 of 690, across 4 files
  • Use Georgia fallback for body textin 39 of 690, across 4 files
  • Maintain text hierarchy and formattingin 39 of 690, across 4 files
  • Use accent colors for non-text shapesin 38 of 690, across 3 files
  • Use RGB values for precise color matchingin 38 of 690, across 3 files
  • Use brand colors for primary text and backgroundsin 36 of 690, across 1 file
  • Read product marketing context file before asking questions, starting, or auditingin 35 of 690, across 23 files
  • Use active voice instead of passive voicein 26 of 690, across 10 files
  • Implement or generate appropriate JSON-LD structured datain 24 of 690, across 17 files
  • Prioritize clarity over clevernessin 22 of 690, across 8 files

Said here and by no other author read

  • use official brand websites as evidence
  • describe promotions visible on the captured page
  • capture at most three screenshots per brand
  • record exact visible claims
  • record evidence URL and timestamp
  • verify each official website URL

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.