Projects
What I'm building and shipping.
OpenCode Dashboard
A self-hosted board for tracking autonomous coding-agent work in real time. It supports task hierarchy, comments, sprint tracking, and analytics so teams can see progress and bottlenecks clearly. Built with Next.js and a local SQLite backend for fast local-first operation.
→github.com/keeeeeeeks/opencode-dashboardAgentPay (APay)
Self-hosted payment infrastructure that gives AI agents their own digital wallets, spending policies, and human-in-the-loop approval flows for on-chain transactions. Includes a real-time admin dashboard for managing per-agent allowlists, policy editing, and audit trails. The policy engine, signing keys, and transaction history never leave your infrastructure.
→github.com/keeeeeeeks/agentpayStoryloom (Liminal)
A platform for publishing, reading, and remixing branching fiction. It supports multiple interactive formats and keeps attribution clear across derivative works. Combines creator tooling with reader-facing discovery so story ecosystems can grow collaboratively.
→storyloom-weld.vercel.appPaperclaws (POMS)
Local-first trading infrastructure combining execution, risk checks, analytics, and a control dashboard. Designed as a composable system where each layer can be improved independently. Focused on disciplined operations, observability, and safer decision-making under live market conditions.
Cookbook (The Roux)
An operations dashboard for evaluating, validating, and managing new product opportunities. Organizes work from idea intake through execution so decisions are traceable across stages. Emphasizes practical scoring, pipeline visibility, and faster iteration loops for product bets.
Clawdina Batch + LLM Gateway
Multi-agent pipeline for research, planning, implementation, and review. Designed for repeatable task execution so results are more consistent than ad-hoc prompting. Combined with the LLM Gateway, both services allow the OpenClaw agent to delegate tasks and be alerted via WebSocket on completion.
Concurrent Craft
Note-first knowledge graph and automated research and publishing pipeline. Uses a SQLite-backed graph store to connect fragments, relationships, and searchable context. Turns messy research notes into structured, publishable outputs with less manual sorting.