Your engineering team, fully autonomous.
Spec in. Secured. Shipped. No standups.
Built at the Deep Agents Hackathon · RSAC 2026 — Shipify is a three-agent AI system that takes a natural-language product spec and ships production-ready, security-reviewed, deployed code entirely on its own. All agents run on OpenClaw.
Airbyte sits at the top as a shared context layer — it syncs Slack conversations and decisions into a vector store (Pinecone), giving all three agents persistent RAG memory. Below that, three agents run in sequence:
- Sean (Kiro) — takes the spec, generates a design doc and task queue via the Kiro CLI, writes the code, and opens a PR
- Christopher (Macroscope) — forks the dev DB, runs a full AST scan on the PR via Macroscope's Slack integration, and either approves or blocks with specific issues
- Amy (TrueFoundry) — promotes the tested DB fork to production, deploys via TrueFoundry, polls health checks, and rolls back instantly via
ghost_pauseif anything fails
Ghost.build is the database layer underneath all three agents — each gets an isolated Postgres DB that follows its own lifecycle: created per task, forked for testing, promoted to prod on approval.
| Agent | Persona | Powered by |
|---|---|---|
| Sean | Developer — methodical, ships PRs while you sleep | Kiro, Ghost.build |
| Christopher | Cybersec / Pentester — red-teams every PR before merge | Macroscope AST |
| Amy | Deployment — zero downtime, zero drama | TrueFoundry, Ghost.build |
| Tool | Role in Shipify |
|---|---|
| Kiro | Spec-driven codegen — prompt → spec → design → tasks → code, hooks fire on every file save |
| Macroscope | AST-based PR review — builds a graph of the codebase, finds hard-to-detect bugs, answers Slack questions grounded in actual code |
| TrueFoundry | Deployment + agent gateway — monitors latency, error rates, token spend per agent, enforces cost budgets in real time |
| Ghost.build | Agent database layer (MCP) — ghost_create, ghost_fork, ghost_promote, ghost_pause give each agent full DB lifecycle control |
| Airbyte | Context storage — 600+ connectors pull Slack history and pipe it into a vector store so all agents share persistent memory |
shipify/
├── main.py # CLI entry point
├── config.py # All env vars
├── orchestrator/pipeline.py # Sean → Christopher → Amy loop
├── agents/
│ ├── developer.py # Kiro: spec → tasks → code → PR
│ ├── tester.py # Macroscope via Slack: AST review
│ └── deployer.py # TrueFoundry: deploy + rollback
├── db/ghost.py # Ghost MCP wrapper
├── context/airbyte.py # Airbyte sync + Pinecone RAG
└── gateway/truefoundry.py # Token tracking + budget enforcement
git clone https://github.com/manufz/RSAC-march27-hackathon
cd RSAC-march27-hackathon
pip install -r requirements.txt
cp .env.example .env # fill in your API keys
python -m shipify.main "Add OAuth2 login with Google" --version 1.4.0The pipeline prints live status from each agent and a per-agent spend report at the end (visible in the TrueFoundry dashboard).
Built at Deep Agents Hackathon · RSAC 2026 · Running on OpenClaw