I'm Gabriel, aka Gabe. Fifteen years working in marketing. The last five years I apply that taste as a Growth Engineer I find what's slowing a team down and build the system that fixes it.
Octant: GTM Engineer & Socials.
Python, SQL, big data, RAG, embeddings, vector databases, MCP, agents, LangGraph, evals, human-in-the-loop, prompt engineering, LLM APIs, TypeScript, JavaScript.
9 packaged AI systems with playbooks, hooks, tests, and a README a stranger can follow. Prospector is open; 8 are private.
Applied AI Software Engineering. MBA, in progress.
Nova. Routes the crew. One inbox. Scribe. Writes in his voice. Interview first. Never posts. Odette. UI, visual meaning. Marlowe. Finds what's missing. Egdod. Architecture before code. Gideon. Why deals stall. Cally. Hunts jobs, applies. Colin. AI video pipeline. Vercel. DNS and deploys. Omnara (Researcher). Sourced analysis. Holloway. Shapes the workflow. Forge. Plans and ships code.
TL;DR & more deets on how I think about work and all things AI in this video, this tweet, this doc, and this one on GTM.
| Project | Stack | What it does |
|---|---|---|
Tixmancer private · main focus |
  |
Agent buys event tickets end to end: 402 offer, signed QR, one-time check-in, receipt on Base; 50 tests. |
Maria private · milestone 1 |
|
Scam-checker Brazilian elders reach on WhatsApp: forward anything suspicious, get a voice verdict in PT-BR. |
| Project | Stack | What it does |
|---|---|---|
| prospector ⭐ |  |
Forkable outbound lead pipeline: two scrapers, Clay enrichment, ICP gates, human approval; ran 100 leads in 48 hours for $4. |
An augment is a folder that teaches an AI assistant a whole job. Plain english: onboarding docs for an AI employee, plus the guardrails that stop it from cutting corners, zipped so anyone can install it.
Every augment here is three parts:
→ Skills = the SOPs. Step-by-step playbooks the AI follows for each task.
→ Hooks = the guardrails. Automatic checks that block the AI the moment it breaks a rule, instead of trusting it to remember.
→ README = the manual. A stranger installs and runs the whole thing w/o me in the room.
These are private rn, no links. Ask me. Prospector up in the GTM section is the open one; fork it.
| Augment | What it does | Access |
|---|---|---|
| ContentOS | 43 playbooks and a 770-chunk voice corpus behind one front door, running my content operation daily, searchable offline. | Private access: follow me and ask for an invite. |
| Dolly | Built this homepage: scroll-driven cinematic sites, machine-verified to actually move before a build passes. | Private access: follow me and ask for an invite. |
| Harrier | Hunts smart-contract and web bounties on Immunefi, Cantina, and Sherlock, and audits my own repos, every finding proven before it counts. | Private access: follow me and ask for an invite. |
| Matt | Plans before code exists, delegates each task, never lets a builder verify its own work, ships only after I pass a quiz on what changed. | Private access: follow me and ask for an invite. |
| Scribe | Decides who types each sentence, me or the machine, before any draft starts; a paired reviewer critiques, never rewrites. | Private access: follow me and ask for an invite. |
| Foundry | Designs apps, UX/UI, and websites through one taste engine, then renders each build in a real browser and iterates before a human looks. | Private access: follow me and ask for an invite. |
| Showrunner | Turns any video ask into a routed production, then measures and watches the finished cut before a human looks. | Private access: follow me and ask for an invite. |
| Masthead | Sends five researchers out on a topic, interviews me on what came back, and tunes the draft for search and AI engines before it reaches my desk. | Private access: follow me and ask for an invite. |
| Project | Lang | What it does |
|---|---|---|
| hedwig-sol ⭐ |  |
Onchain roles for Solana, grant roles not keys: 6 instructions, 21 tests, a documented threat model; Solana Foundation grant. |
| worldcup-pari-market ⭐ | |
Proof-settled World Cup prediction markets on Solana, devnet with 39 Rust tests; demo. |
| kageb ⭐ | |
Private intent pooling for Solana: four real orders in, one aggregate trade out. |
| grimoire | |
The Solana Transaction Doctor: paste a failed signature, get the root cause and fix in plain English, live demo. |
| solana-ship-gate |  |
Pre-deploy safety gate for Solana programs: 4 deterministic checks, blocks unsafe mainnet deploys, MIT licensed. |
| patronus | |
Onchain-attested, runtime-scoped credentials for AI agents on Solana DeFi. |
| worldcup-settlement | |
An AI agent that bets on live World Cup matches on Solana, autonomously. |
| meteora-jup-sol-safety-monitor |  |
Read-only Meteora DLMM PnL and safety monitor with Telegram alerts. |
| solguard | |
Warns Solana users before they get rugged: scores new pump.fun launches on deployer history, liquidity, mint authority, posts risk live. |
| superteam-academy | |
Gamified learning for Solana devs: browser editor, XP, onchain certificates, built for Superteam Brazil. |
| Project | Lang | What it does |
|---|---|---|
| silvia | |
Food-ordering assistant for older adults: chat instead of learning an app, 50+ completed orders. |
| safeskill | |
One-click security audit for your MCP setup: one score, plain English, no CLI required. |
| wingman | |
AI Expert Marketplace: install experts (Design, Code, Marketing) into Claude Code. |
| clawmanship | |
Reviewable decision before a community skill enters a local bundle. |
| agenthub | |
Monad-native AI agent orchestration platform. |
| hermes-aria-theme |  |
Black and white dashboard theme for the Hermes agent: green for success, red for errors, nothing else. |
| Project | Lang | What it does |
|---|---|---|
| walletbrief-monad |  |
Persistent Monad wallet briefs with human-approved, revoke-only EIP-7702 execution. |
| yieldpilot | |
AI copilot for cross-chain yield optimization, built for the Chainlink Convergence Hackathon 2026. |
| Project | Lang | What it does |
|---|---|---|
| beat-claude-engineer-004-analytics-pipeline |  |
Analytics pipeline for the Beat Claude engineer-004 challenge: anomaly detector, test suite, hot-path benchmark. |
| ecommerce-sales-analysis | |
E-commerce sales analysis with Python (pandas, matplotlib) and SQL: RFM segmentation, revenue trends, product performance. |
| solana-narrative-tracker | |
Spots emerging Solana narratives early: tracks 20+ KOLs and dozens of repos, ships build ideas every fortnight. |
| defillama-tvl-anomaly-detector | |
Scans DeFi TVL for what looks wrong: zeroed, -50% in a day, +300% spike, or flat 60 days, catching broken adapters and depegs. |
Also: atelier-gabriel-paz, a landing page for my friend Gabriel Paz, a painter in Botafogo, Rio. No website before; hand-written HTML, no build step.
A high-severity bug in Anchor's LazyAccount code, v0.31.0 to v0.32.1. Lazy::size_of measured a [T; N] array by its first element times N. Every field after an unsized-type array loaded from the wrong byte offset, and exit() wrote the corruption back. Writeup, PoC, fix. Upstream shipped the same fix five weeks later. Bounty paid.
Also built the Solana Vault Standard Extension for a Superteam Brazil bounty. SVS-7 was the round's reviewer pick, then lost to a full-stack entry.
Note
A fund-lockup path in Ern Protocol was also approved through Immunefi. Other reports are still under review. I don't publish details of an open or rejected report; due to program terms
I write about proof, evals, and AI-assisted marketing at my blog.
gabeonchain.com · X @gabe_onchain · LinkedIn