Alex Hormozi–style GTM operation for the LEVEL team and its clients.
A Claude Code plugin with 17 commands, 42 skills (Hormozi frameworks + voice register + output skeletons), 7 specialized subagents, and a gtm-context.md contract that persists across sessions. Hormozi persona always on, brutal voice calibrated by a rubric. Humanizer runs on external copy only.
Quickstart · Commands · Skills · Pipeline · Hooks · Installation · Author
GTM consulting wastes too much time re-explaining the same context to an AI every session. ICP, offer, brand voice, positioning, Core Four split. You spell it all out on Tuesday, lose the details by Friday, spell it out again the next week. Every asset comes out in a slightly different tone than the last. The client notices the inconsistency before you do.
The second problem costs more: AI-generated copy is recognizable in seconds. Em-dash overuse, rule of three, "stands as a testament," "navigating the challenges." When a landing page or ad hook lands on the client's desk reading like that, the project's credibility goes with it.
Hormozi GTM solves both at once. /hormozi-gtm:init runs a guided interview and writes gtm-context.md to the project root; every command after that reads the file automatically, so you never reintroduce anything. Every client-facing output (LP, VSL script, hook, ad, email, proposal) passes through a humanizer agent with EN and PT-BR AI-pattern lists before it's saved, and humanizer_pass: true in the frontmatter is the release gate. The Alex Hormozi persona stays in first person across all 17 commands — no "great question" or "hope that helps" — with a 7-agent pipeline hardened against voice drift and specialists straying out of their lane.
Coverage: acquisition (cold/warm/organic/paid), conversion (LP/VSL/hook/objections), high-ticket closing ($30k+ proposals), delivery (30-day onboarding), retention (churn analysis + winback). Frameworks from all four Hormozi books plus the leaked playbook, with a new strategic layer (niche-selection, founder-market-fit, market-saturation-pivot) that most GTM plugins leave out.
/plugin marketplace add henriquecaner/hormozi-gtm
/plugin install hormozi-gtm@hormozi-gtm-marketplace
/hormozi-gtm:init # 8-question interview → gtm-context.md
/hormozi-gtm:audit # Value Equation diagnostic
/hormozi-gtm:lp --product="product-name" # sales LP with humanizer pass/hormozi-gtm:help # 3-question decision matrix/hormozi-gtm:plan # Core Four split + 90-day money model
/hormozi-gtm:email --type=cold # 5–7 touch sequence
/hormozi-gtm:hooks --n=12 # variants for A/B testing in adsMore scenarios in /hormozi-gtm:help.
17 commands organized by GTM funnel stage:
| Command | Function | Output |
|---|---|---|
/hormozi-gtm:init |
Builds gtm-context.md via a guided interview |
gtm-context.md |
/hormozi-gtm:help |
Decision matrix — recommends a command by objective | (no output) |
/hormozi-gtm:audit |
Value Equation diagnostic (critical bottleneck + top 3 levers) | outputs/audit/ |
| Command | Function | Output |
|---|---|---|
/hormozi-gtm:lp |
Landing page (Grand Slam Offer + 10 structured sections) | outputs/lp/ |
/hormozi-gtm:script |
Long-form 7-step VSL or short-form for Reels/Shorts/TikTok | outputs/script/ |
/hormozi-gtm:hooks |
Battery of 10–20 ad variants (default 15) for A/B testing | outputs/hooks/ |
| Command | Function | Output |
|---|---|---|
/hormozi-gtm:email |
Cold/warm/nurture/re-engagement sequence (5–7 touches) | outputs/email/ |
/hormozi-gtm:objections |
Objection matrix by ICP + sales-call scripts | outputs/objections/ |
/hormozi-gtm:case-study |
Before/after case study + derived assets | outputs/case-studies/ |
/hormozi-gtm:webinar |
B2B 30–45 min structure (educational + sell) | outputs/webinar/ |
/hormozi-gtm:positioning |
Competitive teardown + positioning statement | outputs/positioning/ |
| Command | Function | Output |
|---|---|---|
/hormozi-gtm:pricing |
Review against the 5 Pricing Playbook laws (range + tiering) | outputs/pricing/ |
/hormozi-gtm:plan |
90-day GTM plan (Core Four + 4-tier money model) | outputs/plan/ |
| Command | Function | Output |
|---|---|---|
/hormozi-gtm:client-onboarding |
First 30 days post-sale (5 milestones + triggers) | outputs/client-onboarding/ |
/hormozi-gtm:churn-prevention |
Win/loss + churn analysis + retention playbook (+ optional winback) | outputs/retention/ |
/hormozi-gtm:content-hub |
30–90 day organic content roadmap | outputs/content/ |
| Command | Function | Output |
|---|---|---|
/hormozi-gtm:review |
Brutal feedback + Re-review mode (v1↔v2 delta) | new v{n+1} version |
Argument convention: every command uses --flag=value. Shared flags: --product=<slug>, --ref=<path> (refine), --focus=<section>, --full-rewrite, --no-humanize. Details in CLAUDE.md.
Hormozi frameworks (6)
grand-slam-offer · value-equation · money-models · ltv-cac · core-four · leila-scaling
Strategy (invisible prerequisites) (3)
niche-selection · founder-market-fit · market-saturation-pivot
Copy + Ads (7)
hook-framework · vsl-7-step · ad-copy-formula · scarcity-urgency · guarantees · bonus-stacking · lead-magnets
Advanced acquisition + conversion (4)
content-engine · ad-creative-testing · sales-sequencing · productization
Deliverability + high-ticket closing (2)
email-deliverability · proposal-architecture
Pricing (1)
pricing-playbook
Operational + voice (3)
output-conventions · humanizer-rules · hormozi-voice (brutal voice register + 0–10 brutality rubric)
Output skeletons (16) — internal, loaded by the command via the Skill tool (reliable on both CLI and Cowork)
template-gtm-context · template-lp · template-vsl · template-ad-short · template-hooks-batch · template-email-sequence · template-case-study · template-webinar-agenda · template-content-roadmap · template-churn-analysis · template-review · template-plan · template-pricing-review · template-objections-matrix · template-positioning-map · template-client-onboarding
Skills load only when a command declares them under Active skills. output-conventions and humanizer-rules run on every external output.
The plugin runs as a 7-agent pipeline with structured hand-off contracts, not as a loose pile of prompts. Every command invocation follows:
slash command
↓
hormozi-persona (orchestrator — always)
↓
delegated specialist (by domain)
↓
humanizer (external copy only, before saving)
↓
outputs/<type>/<file>.md
| Agent | Role | When it enters |
|---|---|---|
hormozi-persona |
Orchestrator. Invariant first-person Hormozi voice. Validates each specialist's output before hand-off. | Always — entry point for every command |
offer-architect |
Diagnoses the offer via Value Equation, proposes top 3 levers, rewrites it in one paragraph | /audit, /lp, /script |
ad-architect |
Writes hooks, 7-step VSLs, ad copy | /lp, /script, /hooks, /email, /webinar |
pricing-strategist |
Review against the 5 laws. Range + tiering + anchoring | /pricing, /plan |
leads-strategist |
Core Four split + primary channel + roadmap | /plan, /email |
money-model-architect |
4 tiers (Attraction/Core/Upsell/Continuity) + LTGP/CAC | /plan, /pricing |
humanizer |
Refines copy against AI patterns (PT-BR + EN) | Last step of every external output |
Each specialist declares, in structured Markdown, the exact output format it returns to the orchestrator. Examples:
offer-architect→ Value Equation scores (1–10 per vector) + critical bottleneck + top 3 levers with expected lift + rewritten offerpricing-strategist→ 5 laws scored (green/yellow/red) + $X–Y range + Silver/Gold/Platinum tiering + validation testmoney-model-architect→ 4 tiers with pricing + LTGP/CAC/payback math + text funnel diagram
The orchestrator validates each hand-off before invoking the next agent. If the output doesn't meet the checklist, it sends it back with a specific question instead of improvising.
Each agent has documented behavior for incomplete input:
offer-architectwith no quantitative data → assigns a score at intermediate confidence, marks fields as(estimate)pricing-strategistwith no unit economics → gives a conservative range + flags "market estimate, not a validated recommendation"humanizerrejects the output → saves withhumanizer_pass: false+ a note, orchestrator warns the user
Pipeline details and a 10-step end-to-end example in agents/hormozi-persona.md.
| Hook | Type | Function |
|---|---|---|
session-start |
SessionStart | Banner on session open — reminds you the persona is active and the humanizer is mandatory |
post-tool-aiism-check |
PostToolUse (Write|Edit|MultiEdit) |
Scans freshly modified outputs/*.md for residual AI-isms. Soft warning, never blocks. Defense in depth if the humanizer misses something. |
Scan script in scripts/check-aiisms.py — pure Python stdlib, detects inflated vocabulary, assistant voice, em-dash overuse, hedging. Capped at 20 hits per file to avoid flooding the terminal.
Every command that produces material saves to outputs/<type>/<slug>-{YYYYMMDD}-v{n}.md in the consumer project:
outputs/
├── audit/ ├── email/ ├── retention/
├── lp/ ├── objections/ ├── client-onboarding/
├── script/ ├── case-studies/ ├── content/
├── hooks/ ├── webinar/ └── review/
├── pricing/ ├── positioning/
├── plan/
Versioning increments; it never overwrites without --overwrite. Standard frontmatter: plugin_version, command, version, status, client, product, frameworks (skills used), humanizer_pass, humanizer_mode, voice, language, audit_ref, pricing_ref, parent_version.
Full convention in skills/output-conventions/SKILL.md.
Recommendation: add outputs/ to the consumer project's .gitignore, or version everything depending on your delivery flow.
The Alex Hormozi persona is active across all /hormozi-gtm:* commands. First person, no assistant voice. Even on a plain operational question, the answer comes out in the persona's direct tone. It doesn't relax.
The humanizer runs as the last step before saving any external output (LP, script, hooks, email, case study, webinar, content pieces, winback). It gates external copy only. Diagnostic, strategic, and internal output (audit, review, plan, pricing, objections, positioning, churn analysis, onboarding) stays raw — Hormozi brutal, never softened, no humanizer pass. On external copy the humanizer makes two passes and validates against both EN and PT-BR AI-pattern lists; it unifies the voice and protects the bite (command CTAs, hammer lines, aggressive specificity) instead of sanding it down.
It emits humanizer_pass: <bool> in the output frontmatter. humanizer_pass: false is the external release gate.
The --no-humanize flag exists for debugging and A/B tests, and only on the external-copy commands.
Full rules in skills/humanizer-rules/SKILL.md.
/plugin marketplace add henriquecaner/hormozi-gtm
/plugin install hormozi-gtm@hormozi-gtm-marketplaceDownload the ZIP from the latest release at github.com/henriquecaner/hormozi-gtm/releases/latest. Inside the app:
+→ Create plugin → Upload plugin- Drag in the ZIP
The plugin shows up under Personal plugins and the /hormozi-gtm:* commands become available.
| Environment | Status |
|---|---|
| Claude Code CLI | supported |
| Claude Code Desktop (Mac/Win) | supported |
| Claude Cowork (Desktop app) | supported (ZIP upload) |
| Claude.ai (web) | not supported |
SemVer. Changes are cataloged in CHANGELOG.md. Releases are automated by a v* tag (the release.yml workflow):
- Bump
versionin.claude-plugin/plugin.json+.claude-plugin/marketplace.json(3 places, validated by CI) - Add a
## [X.Y.Z]section toCHANGELOG.md git tag vX.Y.Z && git push origin vX.Y.Z
The workflow fires, builds the ZIP, extracts notes from the CHANGELOG, and publishes the GitHub release.
Longer-horizon roadmap in [0.5.0+] (see CLAUDE.md): customization via settings.json, multiple clients in parallel (gtm-context per slug), an /export command to package outputs.
git clone https://github.com/henriquecaner/hormozi-gtm.git
cd hormozi-gtm
claude --plugin-dir .To build a local ZIP without triggering an official release:
bash scripts/build-zip.shArchitecture details in CLAUDE.md.
LEVEL — Henrique Caner (caner@thelevel.com.br).
Proprietary — LEVEL. Licensed for use by contract. Distribution restricted to the internal team and contractual clients.