Grok BotWorkBack officeAI score · 76/100AI-assisted edit · no obvious prompt-injection riskGitHub

Competitor Radar: Scheduled Monitoring with Diff Summaries

Input competitor list and public pages → scheduled snapshot & diff → push AI summary only when material changes detected

Some experienceStepsWorkflowPromptTip
Who
Small-team operators or product managers tracking 3–10 competitors without time for daily manual browsing
You get
Automatically scan competitor public pages daily, generate structured summaries only when material changes are detected, and push alerts to reduce noise and missed updates
You need

Constraints

Prerequisites: an internet-accessible execution environment (local server, cloud function, or GitHub Actions) capable of scheduled tasks and HTTP requests; a competitor list with names and target page URLs; no historical snapshot needed for the first run. Boundaries and limitations: only scrape publicly accessible pages and respect target sites' robots.txt and terms of service; recommended frequency is once per day—higher frequency may trigger anti-bot measures or unnecessary costs. The source labels this recipe as a fictional example draft, not yet validated on real infrastructure; the underlying model version is unspecified, so actual results may vary by version.

What they did

Source author sit-in provided a tutorial recipe titled "Competitor Radar routine" in the awesome-grokbot-playbooks repository, describing a complete workflow for automatically fetching competitor public pages on a schedule, comparing snapshot diffs, and using AI to generate structured summaries. The recipe includes 5 operational steps, 1 reusable AI prompt template, an expected output format, and 3 known caveats.

How it went

The author reported no actual runtime results. The repository explicitly marks "Validation date: pending (this article is a fictional example draft)" and "Validation level: pending". The output example is illustrative, not a verified result. (author_report)

Worth knowing

Known limitations and gaps: 1) This recipe is labeled by the source author as a fictional example draft and has not been validated on real infrastructure; 2) This site has not built or run the recipe, reproduction=not_tested; 3) The underlying model version is unspecified, so judgment of "meaningless changes" may differ across versions; 4) Page timestamps and dynamic recommendation slots must be filtered beforehand, otherwise they will generate false positives; 5) Only publicly accessible pages should be fetched, respecting robots.txt and keeping frequency restrained; 6) AI summaries must be strictly grounded in diff segments—do not allow free generation, or the model may invent updates that did not occur.

Steps

  1. Set up scheduled trigger

    Configure a scheduled task in your execution environment. Options: cron (Linux/macOS), launchd (macOS), or GitHub Actions. Recommended frequency is once per day to avoid excessive requests that could lead to blocking or unnecessary costs. Verify outbound HTTP requests work before the first scheduled run.

  2. Fetch pages and save snapshots

    Read the competitor list and issue HTTP requests to each competitor's public pages (changelog, pricing page, official social accounts, etc.). Save the returned content as the day's snapshot file. Use a naming convention like '{{date}}-{{competitor-name}}.html' to simplify later comparison.

  3. Compare snapshots and extract diffs

    Compare the current day's snapshot against the previous run's snapshot. Retain only content-level differences, filtering out timestamps, recommendation slots, random ads, and other meaningless changes. If no material differences remain after filtering, terminate the run without proceeding to later steps.

  4. Send diffs to AI for summary

    Feed the filtered diff segments into the AI to obtain a structured summary. Replace variables in the prompt: {{competitor-name}} with the actual competitor name, and the diff segment with the output from Step 3.

    Execution prompt · Source text。AI conversation input; replace {{竞品名}} and diff segment before sending

    以下是竞品 {{竞品名}} 页面今天和上次的差异片段。
用不超过 3 条要点说明发生了什么变化，每条标注类型：
定价 / 功能 / 文案 / 其他。
如果差异只是日期、排版、无意义改动，回复"无实质变化"。
不要推测没有出现在差异里的信息。

  5. Push summary and update snapshot

    If the AI response is not "无实质变化" (no material change), push the summary to your preferred channel (email, IM, or notes app). The output format may follow: 竞品雷达 · {{date}} - Competitor A [Pricing] Entry-tier monthly fee increased; annual discount added - Competitor B [Feature] Changelog mentions "Team Spaces" Other 4 competitors: no material change Finally, overwrite the previous snapshot with the current day's snapshot to serve as the baseline for the next comparison.

Prompt you can copy

Setup guidance · Editorial suggestion。Give to setup assistant to confirm inputs and permissions first

This block is the prompt. Ads are never placed here.

You are setting up a competitor radar routine. Please confirm the following before starting: 1) execution environment type (local / cloud function / GitHub Actions); 2) competitor list (name + URL); 3) scheduling method and frequency; 4) push channel (email / IM / notes); 5) whether a historical snapshot already exists. For first-time setup, leave the snapshot directory empty.

After replacing variables, run a single-day test with one competitor to verify the full chain—fetch, diff, summary, push—before scaling to the full list.

Source

竞品雷达 routine · sit-in · GitHub · 2026-10-17

Comments & feedback

Tried this method? Share results, problems or improvements. Comments appear after moderation. Please avoid sensitive information.

Loading…

    Related