hop.top/<pkg>

Build agent-first command line tools — the 12-Factor AI-CLI specification and conformance Action.

Language-agnostic specification for AI agent conversations

Agent Profile System

IDE-agnostic workspace session manager allowing to start in Claude and resume in Gemini.

Deterministic, isolated, and reproducible multi-branch git worktree wrapper.

Human instructions translated into X-Path capable of finding the intended data even after a page structure or location change.

Instantly transform any agent into a CLI token aware power user that never drifts.

General-purpose benchmarking tool — answers "which approach is better, and by how much?" for any measurable task: tools, implementations, deps, LLM calls, agents.

IDE-agnostic todo list with full syncing with any issue tracking tool for tasks created remotely.

Session, model, and tooling layer on top of any remote compute.

Cross-Assistant Tester — cross-CLI conformance + regression harness for AI-assistant plugins (Claude Code, Gemini, Codex, OpenCode)

Universal package manager for agentskills.io skills.

Deterministic runtime for automating interactive terminal applications and AI CLI agents.

vstar — calendar/vCard semantics library

Protocol-agnostic x402 payment module for agent-native Go applications.

AI model registry CLI — query models.dev with 12-factor agent-safe contracts. Polyglot SDKs: Go, Python, TypeScript, Rust, PHP.

Polyglot toolkit for custom URI schemes (Go, TS, Python, Rust, PHP). Shared contract + parity-tested SDKs.

Write one hook. Run it on every AI coding CLI.

Polyglot AI agent runtime — Go reference runtime + envelope SDKs for TS, Py, Rs, PHP. Implements crtx v0.1 (pronounced 'cortex').

Host-CLI contract for AI-assistant hooks: identity, event maps, envelope and decision shapes, per CLI

Find any coding session across any AI assistant. One command.

Generic multi-channel interaction recorder/replayer with a pluggable adapter interface.

Capability eXecution Router — domain-agnostic dispatch runtime

AGR is a V*-native, temporal, event-sourced, multi-agent orchestration runtime.

Distributed, extensible framework for testing, routing, and validating LLM-based agents.

Self-improvement loop for agent capabilities: evaluate, benchmark, replay

Classification engine — LLM request classification with signal-based routing

Train small advisor models to steer black-box LLMs without fine-tuning.