Open-Source Release: Dual Brand Protection & Anti-Counterfeiting Agent Skills for OpenAI Codex & Google Antigravity
Brand Protection & Anti-Counterfeiting AI Assistant Guide: Codex (brand-protection-skills) × Antigravity (brand-takedown)
Plain-language 5-step operational playbook for two MIT-licensed, zero-dependency IP enforcement Agent Skills by OZ (@zlxmygit): helping non-technical Brand Ops & Legal teams safely onboard brand profiles, capture immutable evidence, screen stolen images, draft Lumen-safe corporate .eml notices, and patrol takedowns without touching code.
1. Built for Non-Technical Brand Ops & Legal: Human-in-the-Loop Safety First
Neither Skill ever sends emails, submits web forms, places test orders, bypasses CAPTCHAs, or auto-adjudicates infringement behind your back. Brand certificates, screenshots, and case records live strictly in private local folders outside any Git repo, while built-in allowlists prevent accidental complaints against your own regional domains or authorized dealers.
For everyday execution, brand-takedown (Antigravity) guides ops teams through 5 plain-language steps—onboard brand profile & public_filing_agent (shielding staff details when providers forward notices to Lumen Database), capture immutable CAP-* snapshots & screen stolen product photos with built-in 64-bit dHash similarity checks, verify official endpoints, approve reviewed drafts to export corporate .eml files, and run daily liveness/RDAP + lookalike/MX radar patrols.
- Zero-Code Daily Control: Ops teams talk directly in natural language using copy-ready prompts; all deterministic Python helpers stay cleanly abstracted behind the agent.
- Privacy & Integrity Guardrails: Optional HMAC-SHA256 evidence manifests prevent tampering, and corporate From-header .eml drafts keep mailbox credentials off disk.
2. Codex Compliance Ledger ($brand-protection) & Multi-Jurisdiction Team Workflow
Ideal when working directly inside OpenAI Codex (installable via a single $skill-installer chat request) or when Legal audits multi-market cases: brand-protection-skills tracks everything inside case-record.json (schema v1.3), strictly separating allegations, timestamped page observations, registry-verified rights, 5-layer provider attributions (ICANN RDAP, DNS/CDN, hosting/SaaS, payment, operator), and 18 action states.
Its read-only followup_queue.py scanner (6h feedback / 24h URL recheck / 72h initial / 168h follow-up defaults, verified by 76 unit tests + 1,120 field mutations) separates routine cooldowns from statutory deadlines, enforces pre-send persistence barriers, and links independent per-channel error retractions via claim_frozen + corrects_action_id whenever new proof clears an authorized dealer.
# 1. Install Codex Edition ($brand-protection) — or just ask $skill-installer in Codex chat git clone https://github.com/zlxmygit/brand-protection-skills.git mkdir -p ~/.agents/skills && cp -R brand-protection-skills/skills/brand-protection ~/.agents/skills/brand-protection python3 -B -m unittest discover -s brand-protection-skills/tests -v # 2. Install Antigravity Edition (brand-takedown) & Run 1-Command Health Check (Python 3.11+) git clone https://github.com/zlxmygit/brand-takedown.git ~/.gemini/config/skills/brand-takedown cd ~/.gemini/config/skills/brand-takedown && python3 scripts/doctor.py python3 scripts/brand_registry.py validate --profile-file examples/sample-profile.json # 3. Daily Patrol, Liveness RDAP Check & Lookalike Phishing Radar python3 scripts/check_liveness.py patrol --slug my-brand --check-rdap python3 scripts/ct_radar.py --slug my-brand --include-lookalikes --lookalike-limit 20
Engineering Comparison Matrix: Anti-Patterns vs. Production Best Practices
| Dimension | Common Anti-Pattern | 2026 Production Standard |
|---|---|---|
| Legal & Privacy Routing | Using copyright DMCA forms for self-shot counterfeits or exposing personal staff emails to Lumen Database | Match notices strictly to verified facts (DMCA for copied assets, Trademark/Design/DSA for counterfeits) & swap in public_filing_agent when public archiving is expected |
| Approval & Sending Handoff | Letting bots auto-fire emails, scrape pages behind CAPTCHAs, or mistake a draft for a real ticket submission | Inspect full notice.txt + attachments yourself, approve digest to export corporate .eml, send via company mail, and log actual ticket receipt |
| Patrols, Counter-Notices & Mistakes | Closing a case on one 404 timeout, repeating chasers on false alarms, or sitting on counter-notices | Verify via liveness + RDAP clienthold probes, freeze & retract across all notified channels on clearance, and escalate counter-notices to legal counsel immediately |
Pre-Flight Production Verification Checklist
- 01Prepare a private local folder outside any Git repo containing brand domains/allowlists, certificates/studio originals, suspect URLs, and a corporate mailbox.
- 02Ask IT or use $skill-installer / doctor.py once to verify your local Python setup returns "ok": true.
- 03Click the toggle button above to open the interactive 5-Step Plain-Language Guide & Situational Ops Prompt Studio.
Found this actionable? Copy standard citation & summary
Formatted with canonical attribution link and executive summary—ready for Slack, Notion, or LLM context.
Put Into Practice: Companion Browser Workbenches
Side-by-Side Document & Prompt Diff Tool
Compare PDF reports, system prompts, and localized copy side by side with zero uploads.
One IP · Purity & AI Node Telemetry
Inspect residential vs. datacenter ASN, fraud risk score, and AI service routing reachability.