Bee Box documentation · directory: https://beebox.run/docs/compared/ · index: https://beebox.run/docs/compared/index.md · root: https://beebox.run/llms.txt Compared: 2026-05-26 — gstack, Garry Tan (source clone studied 2026-05-26 onward; version not recorded); Bee Box at 2026-05-26 Looked for: extensibility mechanism, engineering-process techniques, guardrail enforcement Not looked for: pricing, hosted offerings, community size, personal-assistant features Since then: both projects have changed; treat this page as a snapshot. # Bee Box compared with gstack gstack, published by Garry Tan, is not a personal assistant. It is a collection of about fifty `SKILL.md` files (reusable prompts invoked as slash commands inside Claude Code) for software engineering tasks: planning review, code review, shipping, QA, and security audits. Bee Box is a personal assistant built around a **box**: one directory, kept under version control with a full history of changes (using git), holding **cards** (files with a structured header and a markdown body), for managing a person's ongoing life, not a set of commands a developer invokes per task. The two are different categories of thing; this page compares gstack's process ideas against how Bee Box does the equivalent work, not a competing assistant. **Where they are similar.** Both run on Claude Code, and both rely on loading task-specific instructions on demand rather than keeping everything in one prompt at all times. **Where they differ.** gstack's skills are invoked deliberately, one at a time, by a developer working on a specific piece of code; Bee Box's equivalent, the instructions attached to each kind of card plus procedures, load automatically based on what the agent is touching, with no developer choosing a command. gstack has no data model, memory, connectors, or scheduling of its own; those questions don't apply to it. **What Bee Box borrowed or decided not to.** As of this comparison, the maintainers had adopted gstack's practice of enforcing a destructive-command guardrail through a small piece of Claude Code's own automation rather than a prompt instruction, since it cannot be talked out of running the way a prompt can, and had begun porting a minimal version of its cross-model review skill. Most of gstack's individual skills were read and explicitly set aside as not fitting Bee Box's process, or kept only as a reference point for comparison rather than adopted directly. **What this comparison did not look at:** pricing, hosted offerings, community size, or personal-assistant features, since gstack has none.