Looking for some agents? This is what we do. Agents at your service.
AgentStack Fleet Builder is the public front door for Justin Bellware's agent ecosystem. It helps non-technical users describe what they want to automate, get a plain-English fleet recommendation, and move from the simplest setup path to the best implementation path without having to understand runtimes, harnesses, or infrastructure.
If you want the stronger Codex-native version, the same outcome-first fleets can also be reinforced with the private agi-codex layer for self-learning, self-healing, checkpoints, and visible operating rules.
- Tell the system what you want to build, automate, or grow.
- Get a recommended fleet and a custom plan in natural language, then choose the standard path or the reinforced Codex AGI version.
- Leave your name and email if you want the plan, transcript, and follow-up resources sent to you.
| Fleet | Best For | Public Repos |
|---|---|---|
| SDR Fleet | More meetings, outbound, prospecting, reply handling | autonomous-sdr-agent, linkedin-autopilot-v2, social-sdr |
| GTM Fleet | LinkedIn + outbound + content working together | linkedin-autopilot, linkedin-autopilot-v2, tenxva-gtm |
| Ops / COO Fleet | Dashboards, automations, always-on operators | coo-agent, ops-home-base, business-os-template |
| Builder Fleet | Custom agents, Claude Code/Codex workflows, agent factories | supercode, codex-supercharge-like-claude-code, paperclip-businesses |
| Content Fleet | Content systems, repurposing, YouTube and LinkedIn pipelines | tenxva-dashboard, tenxva-gtm, yt-editor-pipeline |
| Recruiting Agency Fleet | Staffing and recruiting agency systems | recruiterstack, UAIS-Staffing-AI-Vertical, ai-integrators-sales-launch-kit |
Every featured fleet can optionally be paired with the private agi-codex reinforcement layer. That layer adds:
- checked-in
AGENTS.mdpolicy instead of hidden magic - Codex-native hooks and config for safer execution
- repo-local checkpoints and baseline state
- self-healing observations and local-first learning that compound over time
The public funnel stays outcome-first. The reinforced layer is added only when the user wants the stronger self-learning and self-healing version. See docs/agi-codex-reinforcement.md.
Open this repo and start with AGENTS.md. The concierge flow will:
- ask what you're trying to do
- recommend the best fleet
- explain what each repo does
- ask whether you want the standard version or the reinforced Codex AGI version
- walk you through the simplest, guided, and best setup paths
- ask for your contact details only after your plan is ready
The hostable intake app lives in apps/intake. It stores the guided intake, saves the recommendation and transcript, and can sync leads to newsletter and prospecting destinations through webhook-based integrations.
- catalog holds the repo inventory and fleet definitions
- prompts/concierge.md defines the model-facing intake flow
- playbooks contains the simplest, guided, and best paths
- apps/intake is the deployable lead-capture and recommendation app
- scripts/generate-catalog.mjs refreshes the repo catalog from GitHub
- scripts/import-public-repos.mjs vendors public repos into the monorepo
npm run catalog:generate
npm run import:public
npm run devThe public funnel is designed to deliver value first. The lead ask comes after the recommendation:
Want me to send you your custom plan and transcript? Drop your name and email.
The intake app stores:
- the session transcript
- the recommendation snapshot
- the lead record
- the consent record
- sync events for newsletter and prospecting destinations
The first walkthrough script lives in docs/video-script.md. Record a short front-door video and embed it in the repo or a companion landing page.