.prettierignore

# NOTE: `make lint` and `make format` run prettier through
# scripts/prettier.sh, which unions this file with .gitignore. A runtime
# artifact only needs a .gitignore entry -- it does not have to be repeated
# here. The entries below predate that and are kept as-is so `prettier` run
# directly (or through lint-staged) behaves the same way.

_vendor
node_modules
/resources
/assets/css/bundle.css
*/resources
public
.github
.idea
.vscode
.DS_Store
yarn-error.log
.hugo_build.lock
*.md
.devcontainer
assets
config
cypress
data
docfx
infrastructure
theme
tools
layouts
scripts

# Don't check csharp build products
static/programs/*/bin/*
static/programs/*/obj/*

# Program files use hand-written style conventions, not Prettier
static/programs/**/*.ts

# Ignore Python virtual environments
**/.venv/**
**/venv/**

static/registry/
static-prebuilt

# Ignore auto-generated docs
content/docs/reference/pkg

# Ignore config files
tsconfig.json
package.json
package-lock.json
typedoc.json

# Ignore the files we generate during the build and deployment process.
origin-bucket-metadata.json

# Ignore js scripts added to the static folder
static/js

# Conductor workspace state
.context

**/.claude/settings.json
**/.claude/settings.local.json

# Link-checker workflow artifacts (gitignored, generated at runtime)
/.broken-links.json
/.broken-links-pr.txt

# Content-review / docs-review pre-compute artifacts (gitignored, generated at
# runtime). The content-review worker re-runs `make lint` on the fix branch in a
# workspace that still holds these scratch files; without this, `prettier
# --check .` flags them and the re-gate false-fails. This must cover BOTH the
# root-level artifacts AND the `.review-snapshot/` directory (which holds copies
# of the same JSONs, downloaded as a workflow artifact into the publish job) plus
# the fix-scope report — otherwise lint false-fails and the PR is stuck in draft.
/.synthetic.patch
/.fetched-urls.json
/.candidate-claims*.json
/.verified-claims.json
/.frontmatter-validation.json
/.cross-sibling-discovery.json
/.readthrough-findings.json
/.vale-raw.json
/.vale-findings.json
/.content-review-*.json
/.fix-scope-report.json
/.glowup-*.json
/.review-snapshot/
/.pr-body-draft.md
/.lint.log
/.lint-comment.md
/.review-changes.patch
/.review-changes.paths
/.applied.paths
/.review-admin-cache/

# Real-corpus fixture for test_corpus_classifiers.py. Written one claim per
# line on purpose: pretty-printing it is ~5,700 lines of pure noise in every
# diff that regenerates it, which is how a review fixture stops getting read.
scripts/content-review/testdata/