Small, sharp tools focused on one problem: making LLM-driven development reliable enough for production. Specs as contracts, deterministic reviews, context budgets, test gaps — the unglamorous plumbing that turns an AI agent from a demo into something you'd actually ship with.
Evaluates software specifications as formal contracts. Identifies contradictions, hidden assumptions, and missing prerequisites before a line of code is written.
Reviews implementation plans and returns structured critique: contradictions, ambiguities, missing prerequisites, and concrete patches. The check step between spec and code.
Local-first AI code review CLI. Reviews git diffs, commits, ranges, and PRs using Anthropic, OpenAI, Gemini, or Ollama. Emits text, JSON, SARIF, and Markdown — with CI-friendly exit codes and pre-commit hooks.
Intent enforcement for agentic coding systems. Verifies that code matches SPEC.md and PLAN.md — and only that. Catches the drift between what was asked for and what the agent actually built.
Evidence-backed documentation for Go repos. Scans code, builds a Fact Model, then generates architecture / API / data-model / runbook docs via LLM — every claim traceable back to source.
Spec-driven test-strategy analysis. Reads your specs, plans, and Go code to find the tests you don't have — and scaffolds honest skeletons for the ones that matter.
Builds a structural and semantic index of a source repository in SQLite — designed to give AI coding agents a grounded view of the codebase instead of raw text.
Requirements compiler: transforms product ideas into formal, machine-usable specifications for AI coding agents. Uses LLMs as constrained compilers with strict structured outputs — not chatbots.
Deterministic context planner for coding agents. Scores, budgets, and explains which files to load before the agent runs — so LLM context becomes a designed artifact, not a guess.
Terminal-first observability for AI-driven development workflows. The missing "what did the agent actually do?" layer — designed for the command line, not a dashboard.
Autonomous AI agent framework — reliable, observable, tool-using agents with safety controls and deterministic testing.
Graph-based work tracking. Three states, free-form tags, first-class relationships between work items.
Go-native orchestration for stateful, graph-based LLM workflows. Typed state, deterministic replay, checkpointing, and multi-provider adapters.
High-performance vector database in Go, built around LLM memory use cases.
HL7 v2.x parser, encoder, and MLLP network library. Streaming parser, struct marshaling, rule-based validation, ACK/NAK generation.
The original HL7 v2.x decoder/encoder in Go. Widely used across healthcare-IT integrations since 2022.
FHIR client for Go. The interoperability companion to golevel7 — for modern healthcare APIs.
Database visualization tool — PostgreSQL, MySQL, SQLite, SQL Server, MongoDB. Interactive ERD diagrams with schema introspection.
API testing tool — Go backend, React frontend. Collections, env vars with secret encryption, JWT auth, SSRF protection.
Compile-time-configured Claude Code statusline binaries in Go — no jq, no shell, just a single static binary per config.
Symbol-aware semantic code search for Go codebases via MCP. Native AST parsing with domain-driven design support.
Playable Godot ARPG prototype — cursed soul-rings, attunement, loot, echoes — with Go-based content tooling driving the data.
I'm currently available for consulting — systems architecture, developer tooling, LLM infrastructure, and fractional CTO engagements.