# @evolv/flow-lint

Lint Node-RED flow JSON against the EVOLV flow-layout rule
(`.claude/rules/node-red-flow-layout.md`).

## Usage

```bash
# every examples/*.flow.json across all nodes
node tools/flow-lint/bin/flow-lint.js

# one flow
node tools/flow-lint/bin/flow-lint.js nodes/rotatingMachine/examples/01-Basic.flow.json

# CI / JSON output (one line per flow)
node tools/flow-lint/bin/flow-lint.js --json
```

Exit code `1` on any `error`-severity finding; `0` otherwise (warnings do
not fail).

## Rules

| Rule | Severity | Catches |
|---|---|---|
| `UI_XY` | error | `ui-*` nodes with both `x` and `y` at 0 — pile up at canvas origin. |
| `UI_CHART_PROP` | error | `ui-chart` missing one of the required-or-it-renders-blank properties (interpolation, yAxisProperty, …). |
| `UI_CHART_TYPE` | warn | `width` / `height` as string instead of number. |
| `UI_CHART_PALETTE` | warn | `colors` array with fewer than 3 entries. |
| `INJECT_JSON_PROPS` | error | `inject` with `payloadType:"json"` but missing `props[]` — silently treated as string. |
| `LINK_CHANNEL_NAME` | warn | `link out` / `link in` without a `cmd:` / `evt:` / `setup:` prefix. |
| `LINK_BROKEN` | error | `link out` / `link in` referencing a non-existent node id. |
| `LINK_PAIR_TYPE` | error | `link out` paired with anything other than `link in`, or vice versa. |
| `LINK_PAIR_ASYMMETRIC` | error | One side references the other but the partner does not link back. |
| `DEBUG_LOG_IN_FLOW` | warn | `enableLog:"debug"` shipped in a flow — fills container log. |
| `SELF_LOOP` | error | Any node wired to itself. |
| `GROUP_WIDTH_SUM` | warn | `ui-group` widths on a `ui-page` summing to non-multiple of 12 — leaves grid gaps. |

## Why use it

Every rule corresponds to a bug we've actually hit:
- `UI_XY` and `UI_CHART_PROP` — the dashboard pile-up + blank-chart bugs.
- `INJECT_JSON_PROPS` — the silent-string inject bug.
- `LINK_PAIR_ASYMMETRIC` — drift between named channels.
- `SELF_LOOP` — the 250k-msg/s `replace_all` incident.

Run before committing any flow change.