# slhx — AGENTS.md

## Purpose

This file tells coding agents how to work in this repository. It is hand-edited project context, not a generated requirements dump.

## Agent workflow

- Start with requirements before implementation details.
- Read `REQUIREMENTS.md` before changing behavior.
- If behavior changes, update `REQUIREMENTS.md` in the same change.
- Cite relevant requirements in code, tests, or docs as `req: component/001`.
- Run `redgate list`, `redgate refs`, and `redgate health` when requirements change.

## Requirements-first TDD

- Write requirements as intent, not as a dump of current behavior.
- Break requirements into large error classes first: what can go wrong, and what outcome should hold.
- Use tests to pin those error classes before changing code.
- Add adversarial tests for malformed, hostile, ambiguous, missing, duplicated, and boundary inputs.
- Avoid over-codifying existing behavior while the direction is still unclear.
- Add narrower, concrete cases only after requirements converge into a clear design.

## Redgate CLI

- `redgate list` — show requirements as TSV.
- `redgate refs` — show `req:` citations found in the repo.
- `redgate health` — show uncited requirements, duplicate IDs, and stale citations.
- `redgate health --strict` — fail on hard errors: duplicate IDs or stale citations.
- `redgate agents` — print this starter template; review and edit before committing.

## Project commands

- `cargo check --workspace`
- `cargo test --workspace`
- `redgate health --strict`

## Project conventions

- Keep `AGENTS.md` concise; do not paste the requirements catalog into it.
- Requirement IDs use the current form `req: component/001`, not the legacy `req:_component/001` form.
- slhx core stays small: effects, typed ids, registries, and wire schema only.
- Routing, auth, sessions, transport, transitions, sync, and storage belong in integration/user crates.
- Public examples and beginner APIs should use generated resources and `IntoEffect`, not raw ids or runtime opcodes.
- JS runtime changes must preserve root-scoped lookup and avoid selectors, VDOM, expressions, and per-node listeners.