// AI Developer Tools
CLI tools for the AI engineering workflow — cost tracking, prompt ops, context analysis, memory, and agent linting.
agent-lint
Workflow YAML cost estimator + linter 262 tests
promptctl
Claude API toolkit — prompt engineering + code review + doc intelligence 311 tests
context-hygiene
Context window hygiene analyzer for LLM conversations 398 tests
anchormd
AI agent context file generator, auditor + tech debt scanner 693 tests
All 5 Pro Tools — Bundle & Save
anchormd + agent-lint + ai-spend + promptctl + context-hygiene
// Flagship Projects
AI agent framework — autonomous build pipelines, dual-model routing, streaming, MCP server, identity system with guardrails
Fantasy football analytics SaaS — live at benchgoblins.com. Player dossiers, scoring engine, agent pipeline
// Hackathon Projects
EVE Frontier 2026 Hackathon — blockchain intelligence, tribe operations, and on-chain integrity monitoring built on Sui.
WatchTower
LIVERaw blockchain events are noise without analysis
Turns EVE Frontier on-chain Sui events into actionable intelligence — entity dossiers, behavioral fingerprints, earned titles, kill networks, danger zones, and on-chain reputation scoring across 6 dimensions. Reputation flows back to smart contracts for gate enforcement.
Frontier Tribe OS
LIVETribes have no operational infrastructure
Operations platform for EVE Frontier tribes — member roster with SSO, production job board (Kanban), tribe inventory tracking, real-time Sui wallet visibility, and transaction history with on-chain verification. No custodial wallets — client-side signing only via dapp-kit.
Monolith
LIVEBlockchain state anomalies go undetected until launch
Reads EVE Frontier on-chain Sui events and detects state anomalies — orphan objects, supply discrepancies, duplicate mints, stuck objects. 17 deterministic detection rules across 4 checkers generate structured bug reports with on-chain evidence. Discord alerts fire on CRITICAL/HIGH severity.
// Case Studies
Problem
AI coding agents rely on context files (CLAUDE.md, AGENTS.md) for project context — coding standards, architecture, commands, anti-patterns. But these files are written by hand, go stale within days, and nobody audits them. The agent makes worse decisions every time the context drifts from reality.
Solution
Built a CLI that analyzes your codebase and generates accurate context files automatically. It reads pyproject.toml, package.json, Cargo.toml, detects naming conventions by sampling source files, maps architecture trees, extracts commands from CI configs, scans for technical debt, and analyzes GitHub repo health. Then it audits existing files for accuracy and detects behavioral drift across LLM model versions using benchmark suites.
Architecture
- Generator — metadata extraction, pattern analysis, Jinja2 templates
- Auditor — 5 accuracy checkers validate claims against codebase
- Tech Debt Scanner — 9 detection categories, 0-100 scored reports, security/complexity/quality signals
- GitHub Health — repo health analysis, cleanup agent for stale issues/PRs/branches
- Drift Detector — 6 check types, 4 model adapters, YAML benchmark suites, trend visualization
- License Server — FastAPI, SHA-256 hashed keys, rate limiting, activation tracking
Results
BenchGoblins
LIVEFantasy managers flying blind on roster decisions
Fantasy football analytics SaaS with scored LLM routing (signal weights), full ESPN + Yahoo parity for rosters/teams/standings/matchups, structured JSON logging, and the Goblin Verdict engine for data-driven start/sit calls.
EVE Gatekeeper
LIVEEVE Online players need real-time intel to survive
Intel platform with EVE SSO, 13-layer interactive map, wormhole database, Stripe subscriptions, and pilot threat intelligence — threat cards, system summaries, hotzone detection, fleet pilot lookup, and route warnings. PWA for mobile.
Animus
FlagshipAI agents need autonomy with guardrails
Monorepo agent framework with self-improvement loops (self-heal, sandbox, impact scoring, rollback), HybridBackend for Anthropic + Ollama routing, config hot-reload, 37 tools, 6 proactive checks, and a PWA with WebSocket + offline support.
Convergent
PyPIMulti-agent systems produce conflicting outputs
Conflict resolution library with versioned intent graphs, overlap detection, and structured merge strategies. Python + Rust core (PyO3) for performance-critical path resolution across concurrent agent sessions.
Dossier
v0.6.0Document analysis requires cloud dependencies
Local-first document intelligence — NER, classification, OCR, forensics timeline, entity graph analysis, and full-text search (FTS5). Zero cloud dependencies. Everything runs on your machine.
LikX
PyPI + SnapLinux has no screenshot tool with native OCR
GTK3 screenshot and annotation tool with built-in Tesseract OCR, multi-mode capture (region, window, scrolling, GIF), annotation canvas with undo/redo, cloud upload (Imgur, S3, Dropbox), and X11 + Wayland support across GNOME, KDE, and Sway.
G13 Linux
PyPILogitech G13 has no Linux driver
Full-featured Linux driver for the Logitech G13 gameboard — PyQt6 GUI for key mapping, LCD display control, macro recording, per-application profiles, WebSocket API for remote control, and AppImage distribution.