loop-and-gate-build-kit
A judgment layer for building software with an agentic loop — eleven human-decision gates plus the loop-and-gate operating skill.
Open source Open in the app JSON README (API)
About
A judgment layer for building software with an agentic loop — eleven human-decision gates plus the loop-and-gate operating skill.
Details
- Kind
- Plugins
- Topic
- Government & public data
- Publisher
- slogsdon
- Origin
- marketplace
- Category
- ferramentas
- Stars
- 1
- Last push
- 2026-07-27T16:40:13Z
- Repository state
- ativo
- Language
- Shell
- License
- MIT
- Added
- 2026-08-30 01:48:58
- Updated
- 2026-08-30 01:48:58
- Origin id
slogsdon/loop-and-gate-build-kit/loop-and-gate-build-kit
README
# Loop & Gate — Build Kit [](https://www.claudepluginhub.com/plugins/slogsdon-loop-and-gate-build-kit?ref=badge) A judgment layer for building software with an agentic loop. Anyone can install the tools that write code. Almost nobody knows **where to intervene**. This kit is the map of those points — the gates — and how to work each one, even the ones whose expertise you don't have yet. ## The idea in one paragraph An agentic build loop can run brainstorm → plan → build → test → review → ship on its own. It should. But it makes wrong assumptions and runs with them, it overbuilds, it calls a one-off demo "tested," and it can't tell a real problem from a loud one. So the loop has to stop at a fixed set of points and wait for a human. Those points are the gates. Work them well and a mediocre tool stack ships good software. Work them badly and the best tools on the market ship you a fast, tested, well-reviewed mistake. **The gates are the asset. The tools are a disposable appendix.** ## What this is / what it's not **It is** the human-decision gates for an agentic build loop — a runnable gates reference (`reference/gates.md`) plus the `loop-and-gate` skill that runs a change through only the gates it earns. **It is not** the build tooling itself (it sits *on top of* the `superpowers` + `agent-skills` pipeline — the disposable appendix), and it is not an autopilot: every gate stops for your decision. Skip them and you just fail faster with nicer tooling. ## What's in here | Path | What it is | |------|-----------| | `GETTING-STARTED.md` | Zero-assumptions walkthrough — start here if you're new to any of this. | | `reference/gates.md` | The eleven gates in runnable checklist form — the core asset. | | `skills/loop-and-gate/` | The operating skill: run a change through the gates it earns. | | `reference/fan-out.md` | Sub-reference for Gate ∞'s "how wide" call — when running many agents at once earns it, and where the gates sit when it does. | | `reference/pipeline-snapshot.md` | A dated snapshot of the public plugins that fill the space between the gates, with install commands. Labelled disposable on purpose. | | `templates/business-context.md` | A fill-in template for the business-context step (persona, opportunity, KPIs, GTM, risk). | | `.claude-plugin/` | Marketplace + plugin manifest — lets you install this kit via `/plugin marketplace add`. | | `scripts/setup.sh` | Optional convenience for the terminal/clone path. Not needed for plugin install. | ## Install **New to this?** Read **[GETTING-STARTED.md](GETTING-STARTED.md)** — it assumes no coding background and no terminal experience, and gets you from nothing to your first build using the Claude desktop app. **Have Claude Code? Install it as a plugin — no clone, no setup.** In the Code chat (Desktop app, web, or CLI), type: ``` /plugin marketplace add slogsdon/loop-and-gate-build-kit ``` then click **Install** on the menu that appears. The `loop-and-gate` skill is now available: run `/loop-and-gate` at any build decision — it runs **Gate ∞** first to size the change, then walks you through only the gates that change actually needs. You also need the **build pipeline** the gates sit on top of (the gates are the judgment layer, not the loop itself) — two free public plugins, `superpowers` and `agent-skills`. GETTING-STARTED and `reference/pipeline-snapshot.md` list the exact install commands. **Prefer the terminal?** `git clone` this repo and use the `loop-and-gate` skill directly. `scripts/setup.sh` is an optional convenience for that path. ## Usage ``` /loop-and-gate # run any build decision through the gates it earns ``` **Gate ∞** runs first to size the change, then the skill walks you through only the gates that change actually needs — a one-line copy fix earns two gates, a schema migration earns most of them. Worked example below. ## What it looks like A customer reports the lead export is missing a field. You invoke `loop-and-gate`: - **Gate ∞ first.** The skill asks what breaks if this is wrong and whether it's reversible. Export change, no money or auth, reversible → it earns a few gates, not all ten. - **Gate 1.** Who asked, why it matters, what "fixed" means to them. Your call. - **Gate 3.** Approve the plan — and it makes the agent list what it's *not* doing, which surfaces the missing backfill for existing rows. - **Gate 6.** "If this breaks next month, what turns red?" No answer → it's a demo, send it back for a real assertion. - **Gate 9.** Ship to the customer who asked first, tell them, then widen. A one-line copy fix, run through the same skill, earns two gates and ships in minutes. That difference is Gate ∞ doing its job. ## The one rule The kit doesn't decide for you. It surfaces the call, applies the heuristics, puts the agent to work making the decision defensible — then stops. Use it to *work* each gate and you level up on the lens you're missing. Use it to *skip* the gate and you just fail faster, with nicer tooling. ## What this sits next to - **The whole stack** — [Loop & Gate](https://shane.logsdon.io/loop-and-gate/) — the overview: the mental model, all four pieces, and how to install them together. - **The Grow Kit** — [loop-and-gate-grow-kit](https://github.com/slogsdon/loop-and-gate-grow-kit) — is the other half of the loop. This kit takes an idea to shipped software. The Grow Kit takes shipped software to the right people and reads whether it worked. They close a loop through the Foundation vault: the Grow Kit writes market signals where this kit's Gate 0 reads them to decide what to build next. Same method, both halves, one customer at the center. - **The Foundation kit** — [loop-and-gate-foundation](https://github.com/slogsdon/loop-and-gate-foundation) — is the ground this stands on: cross-session memory, capture from your phone, and your voice/taste profiles. If it's present, the operating skill reads `vault/Profiles/` so gate framing matches how you write and judge, and logs each gate decision so your reasoning compounds across sessions. This kit works without it, but they're built to compose — and the Foundation kit is the gentler on-ramp, so **if you're new, start there** (it walks you through installing Claude Code) and add this kit on top. - **Across your devices** — the Foundation's [Working across devices](https://github.com/slogsdon/loop-and-gate-foundation#working-across-devices) guide applies here too. Gate decisions and logs land in the same vault, so the same Remote Control, sync, and worktree discipline runs this kit from your phone, tablet, or desk — nothing kit-specific to set up. - **The field guide** is the narrative version of the gates, with a worked example carried through the whole build of one real product. This kit is the terse, runnable form of the same eleven gates. ## Contributing Issues and pull requests are welcome — see [CONTRIBUTING.md](CONTRIBUTING.md) for how to report bugs, propose gate or skill changes, and the PR conventions. One thing up front: the gates are the asset, so PRs that remove a gate or auto-approve one will be declined. See also the [Code of Conduct](CODE_OF_CONDUCT.md) and the [security policy](.github/SECURITY.md). ## Status The asset (gates reference + operating skill), the business-context template, and the pipeline snapshot with compose-your-own guide are all here. Next: glue to read the Foundation kit's voice/taste profiles, and a final package pass.