Trey Goff Claim this profile

A tiny CLI that gives AI agents a complaint box — log the papercuts they hit during work

Codex autonomous loop controller that prevents premature agent exits with frozen contracts, trusted gates, and fail-closed stop hooks.

Bootstrap repo for autonomous AI-assisted development with Claude Code and Codex CLI

Production-tested long-term memory system for AI agents on OpenClaw. Passive recall, smart compaction, subagent memory, and evaluation harness.

A collection of standalone AI agent skills — image generation, and more

No description.

Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞

All-local Rocket League replay analysis tool for comprehensive performance coaching

Local agent-first CLI. Let Claude or Codex send any model in other harnesses on any task

Agent-first CLI over the full Exa API — 68 operations, one JSON envelope per call, safe by default. Unofficial.

Entity-based memory recall for AI agents. Decay-weighted entity tracking, dynamic assembly, learning feedback loops, and subagent context propagation.

Open-source local desktop pet sidecar for Codex

Local-only, keyword-based PDF redaction. Your documents never leave your machine.

Identity-neutral open-set code archaeology over Bitcoin v0.1

A personal 'as lived in' benchmark: 7 CLI agent harnesses x 3 GPT-5.6 variants x 24 real tasks. Harness, corpus, results, judge scores, and forensics.

The Study — a section of treygoff.com kept by a Claude. Live at claude.treygoff.com

No description.

Local-first, daemon-backed shared memory layer for AI coding harnesses (Claude Code, Codex, Cursor). Design phase.

Homebrew tap for treygoff24's tools (lens, ...)

Source-verified research at function-call latency. Agent-first CLI: every claim comes back with a source URL, quote, and verdict.

Agent-first ElevenLabs CLI — one JSON envelope per command over the full ElevenLabs OpenAPI (320 operations).

Corpus hydration at retina speed — agent-first CLI for fast, cheap, hallucination-firewalled orientation in any directory

Agent-first CLI for natural-language image-library search — caption once with Cerebras vision, search the whole index in one model call. No vector DB.

No description.

Agent harness to make your slop code well-engineered and beautiful.

Versioned belief-management substrate and coordination layer for AI coding agents. Greenfield Rust rewrite of Code Briefcase (V1, Python).

No description.

A menu-bar-only macOS app for DST-aware timezone alignment. SwiftUI, no third-party dependencies.

Agent context runtime for coding agents: hooks, diagnostics, telemetry, and compact code intelligence.

95% token savings. 155x faster queries. 16 languages. LLMs can't read your entire codebase. TLDR extracts structure, traces dependencies, and gives them exactly what they need.

Structural policy engine for TypeScript/JavaScript. Enforces architecture boundaries with deterministic CI output. Tracks rule changes so AI agents can't silently weaken your architecture.

Experimental Memorum-native agent harness prototype.

No description.

Eight-axis multi-agent codebase cleanup for Codex and Claude Code.

Ghostty-based macOS terminal with vertical tabs and notifications for AI coding agents

Simple, token-efficient MCP server for Codex and Claude Code image generation with Nano Banana 2 and Nano Banana Pro.

Opinionated linter configs designed to help AI coding agents write Clean Code

Production-grade context window management for AI agents. Runtime tool-output sanitization, content-addressed artifact storage, context pressure guards, and structured compaction.

A meeting note-taker that talks back.

No description.

Live terminal dashboard for OpenClaw agent sessions — tree view, status icons, transcript viewer. Built with Textual.

No description.

Customize Claude Code's system prompts, create custom toolsets, add custom themes/thinking verbs/spinner, input box style, user message display, sign-in ASCII art, fix bugs (frozen spinner), and add features (context limit customization). Supports both native (binary) and npm installs on all platforms.

What AI coding tools promised. What this one delivers.

No description.

No description.

12-week knee meniscus rehabilitation tracking app

No description.

Controlled A/B experiment evaluating system prompt impact on large language model performance. Shows 6.1pp improvement in factual accuracy using philosophical 'ethics of excellence' framework.