diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
index 4b12adaa456d2..1e181b187adb9 100644
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -259,9 +259,8 @@ Releasing to npm consists of the following phases:
     1. On your local machine, pull from [upstream](https://github.com/puppeteer/puppeteer) and make sure the last commit is the one just merged.
     2. Run `git status` and make sure there are no untracked files.
         - **WHY**: this is to avoid adding unnecessary files to the npm package.
-    3. Run `npm install` to make sure the latest `lib/protocol.d.ts` is generated.
-    4. Run [`npx pkgfiles`](https://www.npmjs.com/package/pkgfiles) to make sure you don't publish anything unnecessary.
-    5. Run `npm publish`. This publishes the `puppeteer` package.
+    3. Run [`npx pkgfiles`](https://www.npmjs.com/package/pkgfiles) to make sure you don't publish anything unnecessary.
+    4. Run `npm publish`. This publishes the `puppeteer` package.
 3. Publish `puppeteer-core` to npm.
     1. Run `./utils/prepare_puppeteer_core.js`. The script changes the name inside `package.json` to `puppeteer-core`.
     2. Run `npm publish`. This publishes the `puppeteer-core` package.
diff --git a/docs/api.md b/docs/api.md
index bcafb5741b04d..7cb55ad832a6c 100644
--- a/docs/api.md
+++ b/docs/api.md
@@ -483,7 +483,7 @@ The default flags that Chromium will be launched with.
 - returns: <[Object]>
 
 Returns a list of devices to be used with [`page.emulate(options)`](#pageemulateoptions). Actual list of
-devices can be found in [lib/DeviceDescriptors.js](https://github.com/puppeteer/puppeteer/blob/master/src/DeviceDescriptors.js).
+devices can be found in [src/DeviceDescriptors.js](https://github.com/puppeteer/puppeteer/blob/master/src/DeviceDescriptors.js).
 
 ```js
 const puppeteer = require('puppeteer');
diff --git a/package.json b/package.json
index 4b54d7375141a..e165913208b79 100644
--- a/package.json
+++ b/package.json
@@ -25,7 +25,7 @@
     "coverage": "cross-env COVERAGE=true npm run unit",
     "tsc": "tsc --version && tsc -p . && cp src/protocol.d.ts lib/ && cp src/externs.d.ts lib/",
     "apply-next-version": "node utils/apply_next_version.js",
-    "bundle": "npx browserify -r ./index.js:puppeteer -o utils/browser/puppeteer-web.js",
+    "bundle": "npm run tsc && npx browserify -r ./index.js:puppeteer -o utils/browser/puppeteer-web.js",
     "test-types": "node utils/doclint/generate_types && tsc --version && tsc -p utils/doclint/generate_types/test/",
     "unit-bundle": "node utils/browser/test.js",
     "update-protocol-d-ts": "node utils/protocol-types-generator"
diff --git a/utils/doclint/README.md b/utils/doclint/README.md
index 46a362d6d2d18..b3f8c366de519 100644
--- a/utils/doclint/README.md
+++ b/utils/doclint/README.md
@@ -6,6 +6,7 @@ Puppeteer's source code.
 Doclint works in a few steps:
 
 1. Read sources in `lib/` folder, parse AST trees and extract public API
+    - note that we run DocLint on the outputted JavaScript in `lib/` rather than the source code in `src/`. We will do this until we have migrated `src/` to be exclusively TypeScript and then we can update DocLint to support TypeScript.
 2. Read sources in `docs/` folder, render markdown to HTML, use puppeteer to traverse the HTML
   and extract described API
 3. Compare one API to another