Open source · Apache 2.0 · v0 — under construction
Strategy-driven
AI agent framework
for ad operations.
Open-source framework for AI agents — Claude Code, Cursor, Codex, Gemini — to operate Google Ads, Meta Ads, Search Console, and GA4. Grounded in your business strategy. Bounded by explicit security rules. Built so a prompt-injection cannot move money.
Three pillars
The properties that distinguish mureo from generic LLM wrappers and AutoGPT-class autonomy.
- a.
Strategy-first
Every analysis consults
STRATEGY.md— persona, USP, brand voice, goals — before it touches metrics. The same CPA spike produces different recommended actions per strategy. - b.
Security-bounded
Credential guard, allow-list rollback gating, GAQL validation, anomaly detection on the action stream. The framework refuses to act outside its allow-list even when instructed to.
- c.
Learnable
A diagnostic knowledge base that gets sharper at your account class over time. Evidence-gated: no recommendation without a sample size, baseline, and confidence threshold.
Quick start
Three commands. Local-first execution — credentials never leave your machine.
$ pip install mureo
$ mureo setup claude-code
# In Claude Code:
> /onboardExplore further
- Manifesto Why we built mureo. The thesis, the three pillars, what mureo is and is not.
- Documentation Install, MCP setup, workflows, security model, architecture.
- Blog Methodology, anti-patterns, case studies, release notes.
- github.com/logly/mureo Source, issues, contribution guidelines.