A local AI agent that researches competitors, plans and safely optimizes Meta Ads campaigns, and creates ad graphics. You get the full source code and docs — no subscription, no vendor lock-in.
A real Meta Ads workflow starts before Ads Manager and ends after the first launch. Research, planning, optimization, and creatives all eat time.
Checking competitors, offers, pricing, reviews, and customer complaints by hand can take days before a campaign even exists.
Losers paused too late, winners scaled too aggressively, and one careless edit can reset Learning Phase.
The same creatives run for weeks. CTR drops, CPM rises, and new angles take more time than they should.
Research in one place, campaign logic in another, graphics somewhere else — and no single flow connecting the data.
$99–$299/month for tools you never fully own. Your data lives in their cloud, and your process depends on their roadmap.
MetaAds Agent connects the whole flow: market research → Meta Ads campaign → ad creatives. Each step feeds the next, so campaigns are built on real market gaps, not guesses.
Scrapes public sources, reviews, pricing pages, marketplaces, forums, and industry portals to find market gaps and repeated customer pains.
Turns the research into a Meta Ads plan: interview, profitability calculator, audience logic, structure, copy angles, and a ready proposal.
Audits campaigns via official Meta Marketing API, flags wasted spend, protects Learning Phase, and suggests safe Stop-Loss or scaling actions.
Generates and edits ad graphics with structured JSON prompting through Gemini or OpenAI — product shots, UGC, before/after, social proof, and more.
Every account-changing action waits for your explicit approval. Campaigns are created PAUSED, and you decide when they go live.
Researches public sources, reviews, pricing, marketplaces, forums, and industry portals to find gaps your campaign can own.
Guided interview → profitability calculator → campaign proposal with structure, budget logic, offers, hooks, and next actions.
Stop-Loss, Winner Scaling, dead element detection, budget allocation recommendations, and Learning Phase safeguards.
10–20 headline & copy variants generated by LLM. Character limits validated. 6 proven hook types.
Verifies CAPI alongside Pixel. Checks event deduplication. Tracks Purchase, Subscribe, Lead events.
Prompt injection protection with 15+ patterns. Secure Keyring (encrypted token storage), token monitoring, audit log, and budget ceiling enforcement.
Zero modifications without your approval. Agent generates a plan — you review and click Proceed.
Supports Meta's unified Advantage+ setup for Sales, App, and Leads — plus manual campaigns when control matters.
Analyzes competitor ads through Meta Ad Library: copy, headlines, platforms, longevity, and angles worth testing.
Generate and edit ad creatives with JSON Prompting via Gemini or OpenAI. Product shots, UGC, lifestyle, before/after, and social proof formats.
One shared brain for Claude, Codex, Cursor, and Antigravity, with workflows and documentation that guide the agent step by step.
Same product, completely different context — generated in seconds with a single prompt.


Background swap + outfit change


Background replacement


Adding elements to scene
python3 -m venv venv
pip install -r requirements.txtTwo commands for the local environment. Python 3.10+.
cp .env.example .env
# Add your Meta token & account IDFill in 5 environment variables from Meta Business Manager.
"Show me results for the last 7 days
and suggest which creatives to refresh"Open in any AI coding tool and chat. That's it.
| Aspect | Typical SaaS $99–$299/mo | MetaAds Agent One-time 497 zł promo |
|---|---|---|
| Annual Cost | $1,188 – $3,588 | One-time 497 zł promo |
| Customization | Limited | Full source code |
| Vendor Lock-in | Yes (data in their cloud) | No — runs on your machine |
| Privacy | Data through 3rd party | Your token, your machine |
| Extensibility | Wait for their roadmap | Add modules yourself |
| Advantage+ Support | Limited control | Full API configuration |
Competition Analyzer — market gaps, reviews, sentiment, PDF reportsMeta Ads Agent — planning, audits, API actions, Stop-Loss, scalingSalvador — AI image generation and editing for ad creativesoptimizer.py — Stop-Loss, scaling, creative fatigueexecutor.py — Human-in-the-Loop gated executioncampaigns.py — campaign drafts and Advantage+ structuresinsights.py — CPA, ROAS, CTR, CPM, frequency, conversionssecurity.py — token handling, audit log, budget limitsOne price. One-time purchase. No subscriptions. No hidden fees. Promo valid until June 7, 2026, 23:59.
No. The agent runs in any AI coding tool — you talk to it in plain English, and it handles the code. One-time .env setup is all that's needed.
Yes. Switch accounts by changing META_AD_ACCOUNT_ID in .env. You can manage as many accounts as you like — one at a time.
No. Every modification requires your explicit approval (Human-in-the-Loop). Campaigns are always created with PAUSED status. You decide when to activate.
The API version is pinned to v25.0. Meta supports each version for ~2 years. The agent uses the official facebook-business Python SDK.
Yes — the agent supports Meta's unified Advantage+ setup for Sales, App, and Leads, plus manual campaigns. It uses campaign budget, Advantage+ audience, and automatic placements instead of legacy campaign creation paths.
This is a digital product delivered immediately after purchase. Refund rules and any required consent for immediate digital delivery are described in the Terms of Service.
No. The agent is built as a set of files (.agent/skills/, .agent/workflows/, Python scripts) — so it works with any AI coding tool that can: ✅ Read instruction files (SKILL.md, workflows), ✅ Run terminal commands (Python scripts), ✅ Follow structured rules. That includes Antigravity, Claude Code, Cursor, Windsurf, Cline, Copilot Workspace, and more.
The agent uses JSON Prompting with Gemini or OpenAI to generate and edit ad creatives. You describe what you need in plain language, and the agent produces a structured JSON prompt for product shots, UGC-style images, lifestyle scenes, before/after creatives, and social proof formats.
No. The landing page shows the core use cases. The package also includes advanced read-only reports for budget allocation, dead elements, performance breakdowns, multi-account dashboards, attribution context after the 2026 measurement changes, landing page checks, Pixel/CAPI verification, screenshot review workflows, token health checks, audit logging, and safety rules for Learning Phase, budget ceilings, and PAUSED campaign drafts.