Skip to content

chore: version examples to 0.1.0 and drop stale .network() doc comment #8

chore: version examples to 0.1.0 and drop stale .network() doc comment

chore: version examples to 0.1.0 and drop stale .network() doc comment #8

Workflow file for this run

name: Lint & Format
# Biome format + lint. INFORMATIONAL for now (continue-on-error), and diff-aware
# on PRs (only files the PR changed). The repo has NOT been reformatted yet —
# doing that days before launch would bury the launch diff. To turn this into a
# blocking gate POST-LAUNCH:
# 1. run `bun run format` (one-time baseline reformat) in its own commit,
# 2. fix/triage any `bun run lint` findings,
# 3. drop the `continue-on-error: true` line below.
# Config + thresholds live in biome.json (noExplicitAny / noNonNullAssertion are
# off there to avoid drowning a large TS codebase in style noise).
on:
pull_request:
push:
branches:
- main
- v*
jobs:
biome:
name: Biome (format + lint)
runs-on: ubuntu-latest
steps:
- name: Checkout
uses: actions/checkout@v4
with:
# Full history so the diff-aware `--since` base ref is reachable.
fetch-depth: 0
- name: Setup Bun
uses: oven-sh/setup-bun@v2
- name: Install dependencies (frozen lockfile)
run: bun ci
- name: Fetch PR base ref
if: github.event_name == 'pull_request'
run: git fetch origin "${{ github.base_ref }}" --depth=1
- name: Biome check (informational)
# continue-on-error keeps this non-blocking until the one-time format
# baseline lands post-launch (see header). Remove it to enforce.
continue-on-error: true
run: |
if [ "${{ github.event_name }}" = "pull_request" ]; then
bunx @biomejs/biome ci --changed --since="origin/${{ github.base_ref }}" .
else
bunx @biomejs/biome ci .
fi
slop:
name: AI-slop ratchet (slop-scan)
# Deterministic detector of AI-generated "slop" patterns (error-obscuring /
# empty catches, log-and-continue, pass-through wrappers, duplicated mocks).
# RATCHET gate on PRs: fails only if the PR ADDS or WORSENS slop vs the base
# — the existing 692-finding baseline never blocks. Existing findings are
# mostly intentional fail-safe patterns (e.g. risk-gate fails CLOSED), so the
# point is to not let NEW slop accrue, not to bulk-fix the baseline. If a
# large refactor trips a false "worsened", add `continue-on-error: true`.
if: github.event_name == 'pull_request'
runs-on: ubuntu-latest
steps:
- name: Checkout
uses: actions/checkout@v4
with:
fetch-depth: 0
- name: Setup Bun
uses: oven-sh/setup-bun@v2
- name: Install dependencies (frozen lockfile)
run: bun ci
- name: Materialize base ref
run: |
git fetch origin "${{ github.base_ref }}" --depth=1
git worktree add ../gordon-base "origin/${{ github.base_ref }}"
- name: slop-scan delta (ratchet)
# Explicit --ignore on both sides so base + head are scoped identically
# regardless of whether the base ref predates slop-scan.config.json.
run: >
bunx slop-scan delta --base ../gordon-base --head .
--ignore 'claude-code-source-code/**'
--ignore 'agents/**'
--ignore 'margin-cli/**'
--ignore 'dist/**'
--ignore '.public-dist/**'
--ignore 'docs/generated/**'
--ignore '**/*.min.js'
--ignore '.site-index.js'
--fail-on added,worsened