agentleFS
Sign inStart free
F0rger123F0rger123/claude-meta-ads-god-mode/AGENTS.md

claude-meta-ads-god-mode

AGENTS.md1 starsChanged 23 days agoMITView on GitHub

Instructions for any AI agent (Claude Code, Codex, Cursor, Gemini CLI, etc.) using or modifying this repository. Asked to set this up, install it, or explain what it is (often after someone pasted the repo link)? Follow docs/SETUP_WIZARD.md: detect where you run, ask the short interview, install, connect accounts read-only first, verify with adsgm doctor / /ads:doctor, then onboard the first client. Everything below applies throughout.

The file

We split this AGENTS.md into the rules it gives an agent

6 rules. About 2 are common, 2 are distinctive, and 2 are very specific to this project.

  • ≈2

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Very specific:

    Read skills/ads-core/SKILL.md first.

    line 11
  2. Very specific:

    All safety state goes through python adsgm.py ...

    line 12
  3. Distinctive:

    Create campaigns, ad sets, ads PAUSED.

    line 13
  4. Common:

    Never print or request credentials.

    line 15
  5. Common:

    No secrets or real client data in commits.

    line 22
  6. Distinctive:

    Keep knowledge labeled and sourced.

    line 24

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

AGENTS.md

Instructions for any AI agent (Claude Code, Codex, Cursor, Gemini CLI, etc.) using or modifying this repository.

0. First thing

Asked to set this up, install it, or explain what it is (often after someone pasted the repo link)? Follow docs/SETUP_WIZARD.md: detect where you run, ask the short interview, install, connect accounts read-only first, verify with adsgm doctor / /ads:doctor, then onboard the first client. Everything below applies throughout.

Using it to manage ads

  1. Read skills/ads-core/SKILL.md first; it defines client resolution, connector routing and the write rules. Then load only the platform skill (skills/meta-ads or skills/google-ads) and the one task skill/workflow the request needs.
  2. All safety state goes through python adsgm.py ... (JSON output). Before any write: adsgm use <client>, verify CLIENT / PLATFORM / ACCOUNT / ENTITY from fresh reads, adsgm plan ..., show the summary, then act.
  3. Create campaigns, ad sets, ads PAUSED. Activation, deletes, tracking changes, geo expansion and out-of-policy budget changes require explicit human approval. Billing and admin changes are never made by an agent.
  4. Hosts without Claude Code hooks do not get the PreToolUse gate automatically. Call python hooks/write_gate.py pre with the hook JSON shape (tool_name, tool_input, permission_mode) before every ad-platform write and obey its decision, or do not enable write-capable connectors.
  5. Never print or request credentials. Never mix clients.
  6. Several clients: one active client at a time (adsgm use). For all-clients work use skills/ads-portfolio (/ads:sweep, /ads:learn, /ads:schedule). Sweeps are read-only. Scheduled runs set ADS_GODMODE_UNATTENDED=1 and are read-only unless a client opts in. Graphic ads: adsgm creative render (local, no paid API); uploads go in PAUSED. Budgets: adsgm budget-plan. Autopilot (per client, off by default; always offer it during onboarding): adsgm autopilot on|off --client X --cap N, workflow workflows/autopilot.md. Niche research: workflows/niche-research.md, adsgm niche ....
  7. Community skills (agrici-ads, marketing-skills, meta-ads-genius and more) ship in bundles/ and community/vendor/ and are chosen on demand through skills/ads-community/SKILL.md (curated picks + catalog.md). Nothing needs installing. python scripts/install.py --codex --native installs them as always-loaded Codex plugins only if the user asks. God Mode rules win; every write goes through adsgm plan and the gate.

Modifying it

  • python tests/run_all.py must pass (stdlib only, no network).
  • No secrets or real client data in commits.
  • Safety logic fails closed; every rule change needs a test.
  • Keep knowledge labeled and sourced; no guaranteed-results claims.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free