Competitor Radar Ads Library
Competitor messaging and swipe-file workflow using Meta Ads Library scraping.
Competitor Radar (Ads Library)
Competitor monitoring, offer tracking, and creative angle mapping from Meta Ads Library.
Know what your competitors are running before your next creative sprint. This skill scrapes the Meta Ads Library via Glued MCP, classifies ads by angle, offer type, CTA, and format, identifies long-running winners and new tests, and generates adaptation ideas with your brand's twist.
What it does
- Validate and scrape — Checks credit/cadence limits, then pulls competitor ads via
scrape_meta_ads_libraryusing either a page ID or keyword search.
- Classify ads — Categorizes each ad by angle, offer type, CTA, format, and platform.
- Identify patterns — Separates persistent ads (likely validated winners) from new tests and detects message shifts over time.
- Generate adaptation ideas — Produces brand-specific creative twists inspired by competitor patterns — never verbatim copies.
- Sync (optional) — Pushes records to Notion and sends alerts to Slack.
Output
output/
├── competitor_swipe.md # Classified ads with patterns, trends, and adaptation ideas
└── competitor_swipe.json # Dated evidence trail with full ad data
What the analysis actually looks like
Each competitor scan includes:
- Ad classification — Angle, offer type, CTA, format, and platform for every ad
- Persistent ads — Long-running creatives flagged as likely validated winners
- New tests — Recently launched ads that signal strategic shifts
- Message shifts — Month-over-month changes in messaging direction
- Adaptation ideas — Brand-specific creative twists (never direct copies)
What this replaces
| Before | After |
|---|---|
| Manually browsing Facebook Ad Library | Automated scraping via Glued MCP |
| No system for tracking competitor ad changes | Month-over-month trend detection |
| Guessing which competitor ads are working | Long-runtime analysis identifies validated winners |
| Copy-pasting competitor ads into a folder | Structured, dated evidence trail in JSON |
| Ad hoc competitor checks before launches | Systematic classification by angle, offer, CTA, and format |
Who this is for
- Creative strategists who want a structured swipe file with adaptation ideas, not just screenshots
- Media buyers tracking competitor offer types and CTA patterns
- Brand managers monitoring competitive messaging shifts over time
- Agency teams running regular competitor audits for multiple clients
Requirements
| Requirement | Details |
|---|---|
| Glued MCP | Access to scrape_meta_ads_library tool |
| Claude Code | Runtime environment (also compatible with Codex) |
| Notion (optional) | For syncing competitor records |
| Slack (optional) | For sending alerts on new competitor ads |
No API keys or environment variables required.
Skill structure
competitor-radar-ads-library/
├── SKILL.md # Skill definition with procedure, quality checks, and guardrails
└── README.md # This file
How it works under the hood
Single-agent skill using one primary Glued MCP tool:
| MCP Tool | Purpose |
|---|---|
scrape_meta_ads_library | Pull competitor ads by page ID or keyword |
Modes: page_id (direct page lookup) or keyword (search-based)
Quality checks: Separates truly new ads from renamed/duplicated creative. Prioritizes long-runtime patterns. Tracks month-over-month angle changes.
Guardrails: Never replicates competitor assets verbatim. Maintains a dated evidence trail. Validates credit and call cadence before requesting data.
For competitive intelligence that drives original creative, not copycat campaigns.