Google Ads CLI for AI agents
Manage Google Ads from OpenClaw, Claude Desktop, or any AI agent that can run shell commands. One CLI, structured JSON, server-side guardrails.
$ npm install -g fullrun
$ fullrun login frun_YOUR_API_KEY
$ fullrun triage
⚡ Account Health Report
─────────────────────────────────────────
🔴 CRITICAL: Ad Groups Missing Ads
2 ad groups have zero active ads — no spend possible.
→ "Brand Terms" (Campaign: High-Intent Keywords)
→ "Competitor Terms" (Campaign: Competitor Capture)
🟠 HIGH: Missing Extensions
3 campaigns are missing sitelinks, callouts, or snippets.
🟡 MEDIUM: Below-First-Page Bids
5 keywords have bids below the first page estimate.How it works
Three steps. Your agent goes from zero to managing Google Ads.
Diagnose
Run fullrun triage to get a prioritized account health report. Issues are ranked from CRITICAL to LOW so the agent knows exactly what to fix first.
$ fullrun triage 🔴 CRITICAL: Budget Mismatch Campaign "Brand Terms" has $0 spend despite $50/day budget. 🟠 HIGH: Poor Ad Strength 4 ads rated POOR by Google. 🟢 Account is 14 days old. Advanced optimization unlocked.
Inspect
Drill into campaigns, keywords, and performance metrics. Every command returns structured data your agent can reason about.
$ fullrun campaigns:list
📊 Campaigns
High-Intent Keywords ● active
Budget: $50.00/day Spend: $142.30
Clicks: 287 Impr: 4,120 CTR: 7.0%
└ Brand Terms 12 keywords · 3 ads
└ Product Terms 8 keywords · 3 ads
Competitor Capture ● active
Budget: $30.00/day Spend: $87.50
Clicks: 156 Impr: 2,890 CTR: 5.4%Optimize
Trigger fullrun run and the AI agent fixes the highest-priority issues automatically. Bids, extensions, negative keywords, ad copy — all handled.
$ fullrun run ✓ Agent run complete ─────────────────────────────── Mode: daily_optimization Actions: 6 → Adjusted bids on 5 keywords → Added 4 sitelinks → Added 8 callouts → Paused keyword "free" (high CPA) → Added 3 negative keywords → Updated ad copy for "Brand Terms"
Built for automation
Everything an AI agent needs to manage Google Ads safely and effectively.
Built for agents
Every command returns structured JSON when piped. Auto-detects TTY for human-readable output in terminals.
Server-side guardrails
Bid caps, budget limits, mutation controls, and PPC best practices enforced on every action. Agents can't break your account.
Triage-first workflow
Start with fullrun triage to get a prioritized health report. Fix the most impactful issues first.
Auto-discoverable
Ships with a SKILL.md file. OpenClaw and Claude auto-discover available commands — no configuration needed.
Full visibility
Campaigns, keywords, performance metrics, and ad strength — all queryable from the command line.
AI-powered optimization
One command triggers a full optimization run — bid adjustments, extensions, negative keywords, ad copy — all autonomous.
Works with your agent
CLI for any agent. MCP for Claude. SKILL.md for auto-discovery.
OpenClaw
Install globally and OpenClaw auto-discovers the CLI via the bundled SKILL.md. Tell your agent to check your ads — it handles the rest.
Claude Desktop & Cursor
Add the MCP server to your config. Ask Claude to check your Google Ads — it calls the Fullrun tools natively.
Any agent
If it can run a shell command, it can use Fullrun. Output is structured JSON when piped — no parsing needed.
Commands
fullrun triageAccount health report with prioritized issuesfullrun campaigns:listAll campaigns with status, budget, and metricsfullrun performance --days 7Clicks, conversions, CPA, ROASfullrun keywords:listKeywords with performance datafullrun runTrigger AI-powered optimization