diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml
index 57538e69..3abde942 100644
--- a/.github/workflows/test.yml
+++ b/.github/workflows/test.yml
@@ -85,8 +85,10 @@ jobs:
       - id: files
         name: Run the action
         uses: ./
+        with:
+          glob-filter: '*'
 
-      - name: Print steps context output
+      - name: Print files
         run: |
           echo 'steps.files.outputs.all=${{ steps.files.outputs.all }}'
           echo 'steps.files.outputs.added=${{ steps.files.outputs.added }}'
@@ -94,3 +96,26 @@ jobs:
           echo 'steps.files.outputs.removed=${{ steps.files.outputs.removed }}'
           echo 'steps.files.outputs.renamed=${{ steps.files.outputs.renamed }}'
           echo 'steps.files.outputs.added_modified=${{ steps.files.outputs.added_modified }}'
+          
+      - id: files-filtered
+        name: Run the action
+        uses: ./
+        with:
+          glob-filter: |
+            *.yml
+            !.github/*/*.yml
+            *.js
+            *.ts
+            package*
+            !*.json
+            
+
+      - name: Print files-filtered
+        run: |
+          echo 'steps.files-filtered.outputs.all=${{ steps.files-filtered.outputs.all }}'
+          echo 'steps.files-filtered.outputs.added=${{ steps.files-filtered.outputs.added }}'
+          echo 'steps.files-filtered.outputs.modified=${{ steps.files-filtered.outputs.modified }}'
+          echo 'steps.files-filtered.outputs.removed=${{ steps.files-filtered.outputs.removed }}'
+          echo 'steps.files-filtered.outputs.renamed=${{ steps.files-filtered.outputs.renamed }}'
+          echo 'steps.files-filtered.outputs.added_modified=${{ steps.files-filtered.outputs.added_modified }}'
+        
diff --git a/README.md b/README.md
index f0433e2b..1c1c858d 100644
--- a/README.md
+++ b/README.md
@@ -1,44 +1,68 @@
+# get-changed-files
+
 [![CI status](https://github.com/Ana06/get-changed-files/workflows/Test/badge.svg)](https://github.com/Ana06/get-changed-files/actions?query=event%3Apush+branch%3Amain)
 [![License](https://img.shields.io/badge/license-MIT-green.svg)](LICENSE.txt)
 
-This project is a fork of https://github.com/jitterbit/get-changed-files, which supports `pull_request_target`, allow to filter files using regular expressions and removes the ahead check.
+This project is a fork of <https://github.com/jitterbit/get-changed-files>, which supports `pull_request_target`, allow to filter files using regular expressions and removes the ahead check.
 
-# Get All Changed Files
+## Get All Changed Files
 
 Get all of the files changed/modified in a pull request (`pull_request` or `pull_request_target`) or push's commits.
 You can choose to get all changed files, only added files, only modified files, only removed files, only renamed files, or all added and modified files.
 These outputs are available via the `steps` output context.
 The `steps` output context exposes the output names `all`, `added`, `modified`, `removed`, `renamed`, and `added_modified`.
 
-# Usage
+## Usage
 
 See [action.yml](action.yml)
 
 ```yaml
-- uses: Ana06/get-changed-files@v1.2
+- uses: Ana06/get-changed-files@v2.0.0
   with:
     # Format of the steps output context.
     # Can be 'space-delimited', 'csv', or 'json'.
     # Default: 'space-delimited'
     format: ''
-    # Filter files using a regex
+    # Filter files using a glob filter
     filter: '*'
 ```
 
-# Scenarios
+## Scenarios
 
-- [Get all changed files as space-delimited](#get-all-changed-files-as-space-delimited)
-- [Get all added and modified files as CSV](#get-all-added-and-modified-files-as-csv)
-- [Get all removed files as JSON](#get-all-removed-files-as-json)
+- [get-changed-files](#get-changed-files)
+  - [Get All Changed Files](#get-all-changed-files)
+  - [Usage](#usage)
+  - [Scenarios](#scenarios)
+    - [Get all changed files as space-delimited](#get-all-changed-files-as-space-delimited)
+    - [Get all changed *.php files as space-delimited](#get-all-changed-php-files-as-space-delimited)
+    - [Get all changed *.yml files but exclude .github/*/*.yml files](#get-all-changed-yml-files-but-exclude-githubyml-files)
+    - [Get all added and modified files as CSV](#get-all-added-and-modified-files-as-csv)
+    - [Get all removed files as JSON](#get-all-removed-files-as-json)
+  - [Install, Build, Lint, Test, and Package](#install-build-lint-test-and-package)
+  - [License](#license)
 
-## Get all changed files as space-delimited
+### Get all changed files as space-delimited
 
 If there are any files with spaces in them, then this method won't work and the step will fail.
 Consider using one of the other formats if that's the case.
 
 ```yaml
 - id: files
-  uses: Ana06/get-changed-files@v1.2
+  uses: Ana06/get-changed-files@v2.0.0
+- run: |
+    for changed_file in ${{ steps.files.outputs.all }}; do
+      echo "Do something with this ${changed_file}."
+    done
+```
+
+### Get all changed *.php files as space-delimited
+
+If there are any files with spaces in them, then this method won't work and the step will fail.
+Consider using one of the other formats if that's the case.
+
+```yaml
+- id: files
+  uses: Ana06/get-changed-files@v2.0.0
   with:
     filter: '*.php'
 - run: |
@@ -47,13 +71,28 @@ Consider using one of the other formats if that's the case.
     done
 ```
 
-## Get all added and modified files as CSV
+### Get all changed *.yml files but exclude .github/*/*.yml files
+
+Be careful that the order of the glob has an importance.
+Therefore, including all YML files first and excluding the YML files of your `.github/*/` directories is the way to go to exclude them.
+If those two globs were inverted, you **would** include all the YML files, with the ones in your `.github/*/` directories.
+
+```yaml
+- uses: Ana06/get-changed-files@v2.0.0
+  with:
+    filter: |
+      *.yml
+      !.github/*/*.yml
+```
+
+### Get all added and modified files as CSV
 
 ```yaml
 - id: files
-  uses: Ana06/get-changed-files@v1.2
+  uses: Ana06/get-changed-files@v2.0.0
   with:
     format: 'csv'
+    filter: '*'
 - run: |
     mapfile -d ',' -t added_modified_files < <(printf '%s,' '${{ steps.files.outputs.added_modified }}')
     for added_modified_file in "${added_modified_files[@]}"; do
@@ -61,13 +100,14 @@ Consider using one of the other formats if that's the case.
     done
 ```
 
-## Get all removed files as JSON
+### Get all removed files as JSON
 
 ```yaml
 - id: files
-  uses: Ana06/get-changed-files@v1.2
+  uses: Ana06/get-changed-files@v2.0.0
   with:
     format: 'json'
+    filter: '*'
 - run: |
     readarray -t removed_files <<<"$(jq -r '.[]' <<<'${{ steps.files.outputs.removed }}')"
     for removed_file in ${removed_files[@]}; do
@@ -75,15 +115,15 @@ Consider using one of the other formats if that's the case.
     done
 ```
 
-# Install, Build, Lint, Test, and Package
+## Install, Build, Lint, Test, and Package
 
 Make sure to do the following before checking in any code changes.
 
 ```bash
-$ yarn
-$ yarn all
+yarn
+yarn all
 ```
 
-# License
+## License
 
 The scripts and documentation in this project are released under the [MIT License](LICENSE)
diff --git a/dist/index.js b/dist/index.js
index 3790d505..32a94da9 100644
--- a/dist/index.js
+++ b/dist/index.js
@@ -1973,6 +1973,33 @@ const windowsRelease = release => {
 module.exports = windowsRelease;
 
 
+/***/ }),
+
+/***/ 82:
+/***/ (function(__unusedmodule, exports) {
+
+"use strict";
+
+// We use any as a valid input type
+/* eslint-disable @typescript-eslint/no-explicit-any */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.toCommandValue = void 0;
+/**
+ * Sanitizes an input into a string so it can be passed into issueCommand safely
+ * @param input input to sanitize into a string
+ */
+function toCommandValue(input) {
+    if (input === null || input === undefined) {
+        return '';
+    }
+    else if (typeof input === 'string' || input instanceof String) {
+        return input;
+    }
+    return JSON.stringify(input);
+}
+exports.toCommandValue = toCommandValue;
+//# sourceMappingURL=utils.js.map
+
 /***/ }),
 
 /***/ 87:
@@ -1982,6 +2009,985 @@ module.exports = require("os");
 
 /***/ }),
 
+/***/ 93:
+/***/ (function(module, __unusedexports, __webpack_require__) {
+
+module.exports = minimatch
+minimatch.Minimatch = Minimatch
+
+var path = { sep: '/' }
+try {
+  path = __webpack_require__(622)
+} catch (er) {}
+
+var GLOBSTAR = minimatch.GLOBSTAR = Minimatch.GLOBSTAR = {}
+var expand = __webpack_require__(306)
+
+var plTypes = {
+  '!': { open: '(?:(?!(?:', close: '))[^/]*?)'},
+  '?': { open: '(?:', close: ')?' },
+  '+': { open: '(?:', close: ')+' },
+  '*': { open: '(?:', close: ')*' },
+  '@': { open: '(?:', close: ')' }
+}
+
+// any single thing other than /
+// don't need to escape / when using new RegExp()
+var qmark = '[^/]'
+
+// * => any number of characters
+var star = qmark + '*?'
+
+// ** when dots are allowed.  Anything goes, except .. and .
+// not (^ or / followed by one or two dots followed by $ or /),
+// followed by anything, any number of times.
+var twoStarDot = '(?:(?!(?:\\\/|^)(?:\\.{1,2})($|\\\/)).)*?'
+
+// not a ^ or / followed by a dot,
+// followed by anything, any number of times.
+var twoStarNoDot = '(?:(?!(?:\\\/|^)\\.).)*?'
+
+// characters that need to be escaped in RegExp.
+var reSpecials = charSet('().*{}+?[]^$\\!')
+
+// "abc" -> { a:true, b:true, c:true }
+function charSet (s) {
+  return s.split('').reduce(function (set, c) {
+    set[c] = true
+    return set
+  }, {})
+}
+
+// normalizes slashes.
+var slashSplit = /\/+/
+
+minimatch.filter = filter
+function filter (pattern, options) {
+  options = options || {}
+  return function (p, i, list) {
+    return minimatch(p, pattern, options)
+  }
+}
+
+function ext (a, b) {
+  a = a || {}
+  b = b || {}
+  var t = {}
+  Object.keys(b).forEach(function (k) {
+    t[k] = b[k]
+  })
+  Object.keys(a).forEach(function (k) {
+    t[k] = a[k]
+  })
+  return t
+}
+
+minimatch.defaults = function (def) {
+  if (!def || !Object.keys(def).length) return minimatch
+
+  var orig = minimatch
+
+  var m = function minimatch (p, pattern, options) {
+    return orig.minimatch(p, pattern, ext(def, options))
+  }
+
+  m.Minimatch = function Minimatch (pattern, options) {
+    return new orig.Minimatch(pattern, ext(def, options))
+  }
+
+  return m
+}
+
+Minimatch.defaults = function (def) {
+  if (!def || !Object.keys(def).length) return Minimatch
+  return minimatch.defaults(def).Minimatch
+}
+
+function minimatch (p, pattern, options) {
+  if (typeof pattern !== 'string') {
+    throw new TypeError('glob pattern string required')
+  }
+
+  if (!options) options = {}
+
+  // shortcut: comments match nothing.
+  if (!options.nocomment && pattern.charAt(0) === '#') {
+    return false
+  }
+
+  // "" only matches ""
+  if (pattern.trim() === '') return p === ''
+
+  return new Minimatch(pattern, options).match(p)
+}
+
+function Minimatch (pattern, options) {
+  if (!(this instanceof Minimatch)) {
+    return new Minimatch(pattern, options)
+  }
+
+  if (typeof pattern !== 'string') {
+    throw new TypeError('glob pattern string required')
+  }
+
+  if (!options) options = {}
+  pattern = pattern.trim()
+
+  // windows support: need to use /, not \
+  if (path.sep !== '/') {
+    pattern = pattern.split(path.sep).join('/')
+  }
+
+  this.options = options
+  this.set = []
+  this.pattern = pattern
+  this.regexp = null
+  this.negate = false
+  this.comment = false
+  this.empty = false
+
+  // make the set of regexps etc.
+  this.make()
+}
+
+Minimatch.prototype.debug = function () {}
+
+Minimatch.prototype.make = make
+function make () {
+  // don't do it more than once.
+  if (this._made) return
+
+  var pattern = this.pattern
+  var options = this.options
+
+  // empty patterns and comments match nothing.
+  if (!options.nocomment && pattern.charAt(0) === '#') {
+    this.comment = true
+    return
+  }
+  if (!pattern) {
+    this.empty = true
+    return
+  }
+
+  // step 1: figure out negation, etc.
+  this.parseNegate()
+
+  // step 2: expand braces
+  var set = this.globSet = this.braceExpand()
+
+  if (options.debug) this.debug = console.error
+
+  this.debug(this.pattern, set)
+
+  // step 3: now we have a set, so turn each one into a series of path-portion
+  // matching patterns.
+  // These will be regexps, except in the case of "**", which is
+  // set to the GLOBSTAR object for globstar behavior,
+  // and will not contain any / characters
+  set = this.globParts = set.map(function (s) {
+    return s.split(slashSplit)
+  })
+
+  this.debug(this.pattern, set)
+
+  // glob --> regexps
+  set = set.map(function (s, si, set) {
+    return s.map(this.parse, this)
+  }, this)
+
+  this.debug(this.pattern, set)
+
+  // filter out everything that didn't compile properly.
+  set = set.filter(function (s) {
+    return s.indexOf(false) === -1
+  })
+
+  this.debug(this.pattern, set)
+
+  this.set = set
+}
+
+Minimatch.prototype.parseNegate = parseNegate
+function parseNegate () {
+  var pattern = this.pattern
+  var negate = false
+  var options = this.options
+  var negateOffset = 0
+
+  if (options.nonegate) return
+
+  for (var i = 0, l = pattern.length
+    ; i < l && pattern.charAt(i) === '!'
+    ; i++) {
+    negate = !negate
+    negateOffset++
+  }
+
+  if (negateOffset) this.pattern = pattern.substr(negateOffset)
+  this.negate = negate
+}
+
+// Brace expansion:
+// a{b,c}d -> abd acd
+// a{b,}c -> abc ac
+// a{0..3}d -> a0d a1d a2d a3d
+// a{b,c{d,e}f}g -> abg acdfg acefg
+// a{b,c}d{e,f}g -> abdeg acdeg abdeg abdfg
+//
+// Invalid sets are not expanded.
+// a{2..}b -> a{2..}b
+// a{b}c -> a{b}c
+minimatch.braceExpand = function (pattern, options) {
+  return braceExpand(pattern, options)
+}
+
+Minimatch.prototype.braceExpand = braceExpand
+
+function braceExpand (pattern, options) {
+  if (!options) {
+    if (this instanceof Minimatch) {
+      options = this.options
+    } else {
+      options = {}
+    }
+  }
+
+  pattern = typeof pattern === 'undefined'
+    ? this.pattern : pattern
+
+  if (typeof pattern === 'undefined') {
+    throw new TypeError('undefined pattern')
+  }
+
+  if (options.nobrace ||
+    !pattern.match(/\{.*\}/)) {
+    // shortcut. no need to expand.
+    return [pattern]
+  }
+
+  return expand(pattern)
+}
+
+// parse a component of the expanded set.
+// At this point, no pattern may contain "/" in it
+// so we're going to return a 2d array, where each entry is the full
+// pattern, split on '/', and then turned into a regular expression.
+// A regexp is made at the end which joins each array with an
+// escaped /, and another full one which joins each regexp with |.
+//
+// Following the lead of Bash 4.1, note that "**" only has special meaning
+// when it is the *only* thing in a path portion.  Otherwise, any series
+// of * is equivalent to a single *.  Globstar behavior is enabled by
+// default, and can be disabled by setting options.noglobstar.
+Minimatch.prototype.parse = parse
+var SUBPARSE = {}
+function parse (pattern, isSub) {
+  if (pattern.length > 1024 * 64) {
+    throw new TypeError('pattern is too long')
+  }
+
+  var options = this.options
+
+  // shortcuts
+  if (!options.noglobstar && pattern === '**') return GLOBSTAR
+  if (pattern === '') return ''
+
+  var re = ''
+  var hasMagic = !!options.nocase
+  var escaping = false
+  // ? => one single character
+  var patternListStack = []
+  var negativeLists = []
+  var stateChar
+  var inClass = false
+  var reClassStart = -1
+  var classStart = -1
+  // . and .. never match anything that doesn't start with .,
+  // even when options.dot is set.
+  var patternStart = pattern.charAt(0) === '.' ? '' // anything
+  // not (start or / followed by . or .. followed by / or end)
+  : options.dot ? '(?!(?:^|\\\/)\\.{1,2}(?:$|\\\/))'
+  : '(?!\\.)'
+  var self = this
+
+  function clearStateChar () {
+    if (stateChar) {
+      // we had some state-tracking character
+      // that wasn't consumed by this pass.
+      switch (stateChar) {
+        case '*':
+          re += star
+          hasMagic = true
+        break
+        case '?':
+          re += qmark
+          hasMagic = true
+        break
+        default:
+          re += '\\' + stateChar
+        break
+      }
+      self.debug('clearStateChar %j %j', stateChar, re)
+      stateChar = false
+    }
+  }
+
+  for (var i = 0, len = pattern.length, c
+    ; (i < len) && (c = pattern.charAt(i))
+    ; i++) {
+    this.debug('%s\t%s %s %j', pattern, i, re, c)
+
+    // skip over any that are escaped.
+    if (escaping && reSpecials[c]) {
+      re += '\\' + c
+      escaping = false
+      continue
+    }
+
+    switch (c) {
+      case '/':
+        // completely not allowed, even escaped.
+        // Should already be path-split by now.
+        return false
+
+      case '\\':
+        clearStateChar()
+        escaping = true
+      continue
+
+      // the various stateChar values
+      // for the "extglob" stuff.
+      case '?':
+      case '*':
+      case '+':
+      case '@':
+      case '!':
+        this.debug('%s\t%s %s %j <-- stateChar', pattern, i, re, c)
+
+        // all of those are literals inside a class, except that
+        // the glob [!a] means [^a] in regexp
+        if (inClass) {
+          this.debug('  in class')
+          if (c === '!' && i === classStart + 1) c = '^'
+          re += c
+          continue
+        }
+
+        // if we already have a stateChar, then it means
+        // that there was something like ** or +? in there.
+        // Handle the stateChar, then proceed with this one.
+        self.debug('call clearStateChar %j', stateChar)
+        clearStateChar()
+        stateChar = c
+        // if extglob is disabled, then +(asdf|foo) isn't a thing.
+        // just clear the statechar *now*, rather than even diving into
+        // the patternList stuff.
+        if (options.noext) clearStateChar()
+      continue
+
+      case '(':
+        if (inClass) {
+          re += '('
+          continue
+        }
+
+        if (!stateChar) {
+          re += '\\('
+          continue
+        }
+
+        patternListStack.push({
+          type: stateChar,
+          start: i - 1,
+          reStart: re.length,
+          open: plTypes[stateChar].open,
+          close: plTypes[stateChar].close
+        })
+        // negation is (?:(?!js)[^/]*)
+        re += stateChar === '!' ? '(?:(?!(?:' : '(?:'
+        this.debug('plType %j %j', stateChar, re)
+        stateChar = false
+      continue
+
+      case ')':
+        if (inClass || !patternListStack.length) {
+          re += '\\)'
+          continue
+        }
+
+        clearStateChar()
+        hasMagic = true
+        var pl = patternListStack.pop()
+        // negation is (?:(?!js)[^/]*)
+        // The others are (?:<pattern>)<type>
+        re += pl.close
+        if (pl.type === '!') {
+          negativeLists.push(pl)
+        }
+        pl.reEnd = re.length
+      continue
+
+      case '|':
+        if (inClass || !patternListStack.length || escaping) {
+          re += '\\|'
+          escaping = false
+          continue
+        }
+
+        clearStateChar()
+        re += '|'
+      continue
+
+      // these are mostly the same in regexp and glob
+      case '[':
+        // swallow any state-tracking char before the [
+        clearStateChar()
+
+        if (inClass) {
+          re += '\\' + c
+          continue
+        }
+
+        inClass = true
+        classStart = i
+        reClassStart = re.length
+        re += c
+      continue
+
+      case ']':
+        //  a right bracket shall lose its special
+        //  meaning and represent itself in
+        //  a bracket expression if it occurs
+        //  first in the list.  -- POSIX.2 2.8.3.2
+        if (i === classStart + 1 || !inClass) {
+          re += '\\' + c
+          escaping = false
+          continue
+        }
+
+        // handle the case where we left a class open.
+        // "[z-a]" is valid, equivalent to "\[z-a\]"
+        if (inClass) {
+          // split where the last [ was, make sure we don't have
+          // an invalid re. if so, re-walk the contents of the
+          // would-be class to re-translate any characters that
+          // were passed through as-is
+          // TODO: It would probably be faster to determine this
+          // without a try/catch and a new RegExp, but it's tricky
+          // to do safely.  For now, this is safe and works.
+          var cs = pattern.substring(classStart + 1, i)
+          try {
+            RegExp('[' + cs + ']')
+          } catch (er) {
+            // not a valid class!
+            var sp = this.parse(cs, SUBPARSE)
+            re = re.substr(0, reClassStart) + '\\[' + sp[0] + '\\]'
+            hasMagic = hasMagic || sp[1]
+            inClass = false
+            continue
+          }
+        }
+
+        // finish up the class.
+        hasMagic = true
+        inClass = false
+        re += c
+      continue
+
+      default:
+        // swallow any state char that wasn't consumed
+        clearStateChar()
+
+        if (escaping) {
+          // no need
+          escaping = false
+        } else if (reSpecials[c]
+          && !(c === '^' && inClass)) {
+          re += '\\'
+        }
+
+        re += c
+
+    } // switch
+  } // for
+
+  // handle the case where we left a class open.
+  // "[abc" is valid, equivalent to "\[abc"
+  if (inClass) {
+    // split where the last [ was, and escape it
+    // this is a huge pita.  We now have to re-walk
+    // the contents of the would-be class to re-translate
+    // any characters that were passed through as-is
+    cs = pattern.substr(classStart + 1)
+    sp = this.parse(cs, SUBPARSE)
+    re = re.substr(0, reClassStart) + '\\[' + sp[0]
+    hasMagic = hasMagic || sp[1]
+  }
+
+  // handle the case where we had a +( thing at the *end*
+  // of the pattern.
+  // each pattern list stack adds 3 chars, and we need to go through
+  // and escape any | chars that were passed through as-is for the regexp.
+  // Go through and escape them, taking care not to double-escape any
+  // | chars that were already escaped.
+  for (pl = patternListStack.pop(); pl; pl = patternListStack.pop()) {
+    var tail = re.slice(pl.reStart + pl.open.length)
+    this.debug('setting tail', re, pl)
+    // maybe some even number of \, then maybe 1 \, followed by a |
+    tail = tail.replace(/((?:\\{2}){0,64})(\\?)\|/g, function (_, $1, $2) {
+      if (!$2) {
+        // the | isn't already escaped, so escape it.
+        $2 = '\\'
+      }
+
+      // need to escape all those slashes *again*, without escaping the
+      // one that we need for escaping the | character.  As it works out,
+      // escaping an even number of slashes can be done by simply repeating
+      // it exactly after itself.  That's why this trick works.
+      //
+      // I am sorry that you have to see this.
+      return $1 + $1 + $2 + '|'
+    })
+
+    this.debug('tail=%j\n   %s', tail, tail, pl, re)
+    var t = pl.type === '*' ? star
+      : pl.type === '?' ? qmark
+      : '\\' + pl.type
+
+    hasMagic = true
+    re = re.slice(0, pl.reStart) + t + '\\(' + tail
+  }
+
+  // handle trailing things that only matter at the very end.
+  clearStateChar()
+  if (escaping) {
+    // trailing \\
+    re += '\\\\'
+  }
+
+  // only need to apply the nodot start if the re starts with
+  // something that could conceivably capture a dot
+  var addPatternStart = false
+  switch (re.charAt(0)) {
+    case '.':
+    case '[':
+    case '(': addPatternStart = true
+  }
+
+  // Hack to work around lack of negative lookbehind in JS
+  // A pattern like: *.!(x).!(y|z) needs to ensure that a name
+  // like 'a.xyz.yz' doesn't match.  So, the first negative
+  // lookahead, has to look ALL the way ahead, to the end of
+  // the pattern.
+  for (var n = negativeLists.length - 1; n > -1; n--) {
+    var nl = negativeLists[n]
+
+    var nlBefore = re.slice(0, nl.reStart)
+    var nlFirst = re.slice(nl.reStart, nl.reEnd - 8)
+    var nlLast = re.slice(nl.reEnd - 8, nl.reEnd)
+    var nlAfter = re.slice(nl.reEnd)
+
+    nlLast += nlAfter
+
+    // Handle nested stuff like *(*.js|!(*.json)), where open parens
+    // mean that we should *not* include the ) in the bit that is considered
+    // "after" the negated section.
+    var openParensBefore = nlBefore.split('(').length - 1
+    var cleanAfter = nlAfter
+    for (i = 0; i < openParensBefore; i++) {
+      cleanAfter = cleanAfter.replace(/\)[+*?]?/, '')
+    }
+    nlAfter = cleanAfter
+
+    var dollar = ''
+    if (nlAfter === '' && isSub !== SUBPARSE) {
+      dollar = '$'
+    }
+    var newRe = nlBefore + nlFirst + nlAfter + dollar + nlLast
+    re = newRe
+  }
+
+  // if the re is not "" at this point, then we need to make sure
+  // it doesn't match against an empty path part.
+  // Otherwise a/* will match a/, which it should not.
+  if (re !== '' && hasMagic) {
+    re = '(?=.)' + re
+  }
+
+  if (addPatternStart) {
+    re = patternStart + re
+  }
+
+  // parsing just a piece of a larger pattern.
+  if (isSub === SUBPARSE) {
+    return [re, hasMagic]
+  }
+
+  // skip the regexp for non-magical patterns
+  // unescape anything in it, though, so that it'll be
+  // an exact match against a file etc.
+  if (!hasMagic) {
+    return globUnescape(pattern)
+  }
+
+  var flags = options.nocase ? 'i' : ''
+  try {
+    var regExp = new RegExp('^' + re + '$', flags)
+  } catch (er) {
+    // If it was an invalid regular expression, then it can't match
+    // anything.  This trick looks for a character after the end of
+    // the string, which is of course impossible, except in multi-line
+    // mode, but it's not a /m regex.
+    return new RegExp('$.')
+  }
+
+  regExp._glob = pattern
+  regExp._src = re
+
+  return regExp
+}
+
+minimatch.makeRe = function (pattern, options) {
+  return new Minimatch(pattern, options || {}).makeRe()
+}
+
+Minimatch.prototype.makeRe = makeRe
+function makeRe () {
+  if (this.regexp || this.regexp === false) return this.regexp
+
+  // at this point, this.set is a 2d array of partial
+  // pattern strings, or "**".
+  //
+  // It's better to use .match().  This function shouldn't
+  // be used, really, but it's pretty convenient sometimes,
+  // when you just want to work with a regex.
+  var set = this.set
+
+  if (!set.length) {
+    this.regexp = false
+    return this.regexp
+  }
+  var options = this.options
+
+  var twoStar = options.noglobstar ? star
+    : options.dot ? twoStarDot
+    : twoStarNoDot
+  var flags = options.nocase ? 'i' : ''
+
+  var re = set.map(function (pattern) {
+    return pattern.map(function (p) {
+      return (p === GLOBSTAR) ? twoStar
+      : (typeof p === 'string') ? regExpEscape(p)
+      : p._src
+    }).join('\\\/')
+  }).join('|')
+
+  // must match entire pattern
+  // ending in a * or ** will make it less strict.
+  re = '^(?:' + re + ')$'
+
+  // can match anything, as long as it's not this.
+  if (this.negate) re = '^(?!' + re + ').*$'
+
+  try {
+    this.regexp = new RegExp(re, flags)
+  } catch (ex) {
+    this.regexp = false
+  }
+  return this.regexp
+}
+
+minimatch.match = function (list, pattern, options) {
+  options = options || {}
+  var mm = new Minimatch(pattern, options)
+  list = list.filter(function (f) {
+    return mm.match(f)
+  })
+  if (mm.options.nonull && !list.length) {
+    list.push(pattern)
+  }
+  return list
+}
+
+Minimatch.prototype.match = match
+function match (f, partial) {
+  this.debug('match', f, this.pattern)
+  // short-circuit in the case of busted things.
+  // comments, etc.
+  if (this.comment) return false
+  if (this.empty) return f === ''
+
+  if (f === '/' && partial) return true
+
+  var options = this.options
+
+  // windows: need to use /, not \
+  if (path.sep !== '/') {
+    f = f.split(path.sep).join('/')
+  }
+
+  // treat the test path as a set of pathparts.
+  f = f.split(slashSplit)
+  this.debug(this.pattern, 'split', f)
+
+  // just ONE of the pattern sets in this.set needs to match
+  // in order for it to be valid.  If negating, then just one
+  // match means that we have failed.
+  // Either way, return on the first hit.
+
+  var set = this.set
+  this.debug(this.pattern, 'set', set)
+
+  // Find the basename of the path by looking for the last non-empty segment
+  var filename
+  var i
+  for (i = f.length - 1; i >= 0; i--) {
+    filename = f[i]
+    if (filename) break
+  }
+
+  for (i = 0; i < set.length; i++) {
+    var pattern = set[i]
+    var file = f
+    if (options.matchBase && pattern.length === 1) {
+      file = [filename]
+    }
+    var hit = this.matchOne(file, pattern, partial)
+    if (hit) {
+      if (options.flipNegate) return true
+      return !this.negate
+    }
+  }
+
+  // didn't get any hits.  this is success if it's a negative
+  // pattern, failure otherwise.
+  if (options.flipNegate) return false
+  return this.negate
+}
+
+// set partial to true to test if, for example,
+// "/a/b" matches the start of "/*/b/*/d"
+// Partial means, if you run out of file before you run
+// out of pattern, then that's fine, as long as all
+// the parts match.
+Minimatch.prototype.matchOne = function (file, pattern, partial) {
+  var options = this.options
+
+  this.debug('matchOne',
+    { 'this': this, file: file, pattern: pattern })
+
+  this.debug('matchOne', file.length, pattern.length)
+
+  for (var fi = 0,
+      pi = 0,
+      fl = file.length,
+      pl = pattern.length
+      ; (fi < fl) && (pi < pl)
+      ; fi++, pi++) {
+    this.debug('matchOne loop')
+    var p = pattern[pi]
+    var f = file[fi]
+
+    this.debug(pattern, p, f)
+
+    // should be impossible.
+    // some invalid regexp stuff in the set.
+    if (p === false) return false
+
+    if (p === GLOBSTAR) {
+      this.debug('GLOBSTAR', [pattern, p, f])
+
+      // "**"
+      // a/**/b/**/c would match the following:
+      // a/b/x/y/z/c
+      // a/x/y/z/b/c
+      // a/b/x/b/x/c
+      // a/b/c
+      // To do this, take the rest of the pattern after
+      // the **, and see if it would match the file remainder.
+      // If so, return success.
+      // If not, the ** "swallows" a segment, and try again.
+      // This is recursively awful.
+      //
+      // a/**/b/**/c matching a/b/x/y/z/c
+      // - a matches a
+      // - doublestar
+      //   - matchOne(b/x/y/z/c, b/**/c)
+      //     - b matches b
+      //     - doublestar
+      //       - matchOne(x/y/z/c, c) -> no
+      //       - matchOne(y/z/c, c) -> no
+      //       - matchOne(z/c, c) -> no
+      //       - matchOne(c, c) yes, hit
+      var fr = fi
+      var pr = pi + 1
+      if (pr === pl) {
+        this.debug('** at the end')
+        // a ** at the end will just swallow the rest.
+        // We have found a match.
+        // however, it will not swallow /.x, unless
+        // options.dot is set.
+        // . and .. are *never* matched by **, for explosively
+        // exponential reasons.
+        for (; fi < fl; fi++) {
+          if (file[fi] === '.' || file[fi] === '..' ||
+            (!options.dot && file[fi].charAt(0) === '.')) return false
+        }
+        return true
+      }
+
+      // ok, let's see if we can swallow whatever we can.
+      while (fr < fl) {
+        var swallowee = file[fr]
+
+        this.debug('\nglobstar while', file, fr, pattern, pr, swallowee)
+
+        // XXX remove this slice.  Just pass the start index.
+        if (this.matchOne(file.slice(fr), pattern.slice(pr), partial)) {
+          this.debug('globstar found match!', fr, fl, swallowee)
+          // found a match.
+          return true
+        } else {
+          // can't swallow "." or ".." ever.
+          // can only swallow ".foo" when explicitly asked.
+          if (swallowee === '.' || swallowee === '..' ||
+            (!options.dot && swallowee.charAt(0) === '.')) {
+            this.debug('dot detected!', file, fr, pattern, pr)
+            break
+          }
+
+          // ** swallows a segment, and continue.
+          this.debug('globstar swallow a segment, and continue')
+          fr++
+        }
+      }
+
+      // no match was found.
+      // However, in partial mode, we can't say this is necessarily over.
+      // If there's more *pattern* left, then
+      if (partial) {
+        // ran out of file
+        this.debug('\n>>> no match, partial?', file, fr, pattern, pr)
+        if (fr === fl) return true
+      }
+      return false
+    }
+
+    // something other than **
+    // non-magic patterns just have to match exactly
+    // patterns with magic have been turned into regexps.
+    var hit
+    if (typeof p === 'string') {
+      if (options.nocase) {
+        hit = f.toLowerCase() === p.toLowerCase()
+      } else {
+        hit = f === p
+      }
+      this.debug('string match', p, f, hit)
+    } else {
+      hit = f.match(p)
+      this.debug('pattern match', p, f, hit)
+    }
+
+    if (!hit) return false
+  }
+
+  // Note: ending in / means that we'll get a final ""
+  // at the end of the pattern.  This can only match a
+  // corresponding "" at the end of the file.
+  // If the file ends in /, then it can only match a
+  // a pattern that ends in /, unless the pattern just
+  // doesn't have any more for it. But, a/b/ should *not*
+  // match "a/b/*", even though "" matches against the
+  // [^/]*? pattern, except in partial mode, where it might
+  // simply not be reached yet.
+  // However, a/b/ should still satisfy a/*
+
+  // now either we fell off the end of the pattern, or we're done.
+  if (fi === fl && pi === pl) {
+    // ran out of pattern and filename at the same time.
+    // an exact hit!
+    return true
+  } else if (fi === fl) {
+    // ran out of file, but still had pattern left.
+    // this is ok if we're doing the match as part of
+    // a glob fs traversal.
+    return partial
+  } else if (pi === pl) {
+    // ran out of pattern, still have file left.
+    // this is only acceptable if we're on the very last
+    // empty segment of a file with a trailing slash.
+    // a/* should match a/b/
+    var emptyFileEnd = (fi === fl - 1) && (file[fi] === '')
+    return emptyFileEnd
+  }
+
+  // should be unreachable.
+  throw new Error('wtf?')
+}
+
+// replace stuff like \* with *
+function globUnescape (s) {
+  return s.replace(/\\(.)/g, '$1')
+}
+
+function regExpEscape (s) {
+  return s.replace(/[-[\]{}()*+?.,\\^$|#\s]/g, '\\$&')
+}
+
+
+/***/ }),
+
+/***/ 102:
+/***/ (function(__unusedmodule, exports, __webpack_require__) {
+
+"use strict";
+
+// For internal use, subject to change.
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    Object.defineProperty(o, k2, { enumerable: true, get: function() { return m[k]; } });
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
+    Object.defineProperty(o, "default", { enumerable: true, value: v });
+}) : function(o, v) {
+    o["default"] = v;
+});
+var __importStar = (this && this.__importStar) || function (mod) {
+    if (mod && mod.__esModule) return mod;
+    var result = {};
+    if (mod != null) for (var k in mod) if (k !== "default" && Object.hasOwnProperty.call(mod, k)) __createBinding(result, mod, k);
+    __setModuleDefault(result, mod);
+    return result;
+};
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.issueCommand = void 0;
+// We use any as a valid input type
+/* eslint-disable @typescript-eslint/no-explicit-any */
+const fs = __importStar(__webpack_require__(747));
+const os = __importStar(__webpack_require__(87));
+const utils_1 = __webpack_require__(82);
+function issueCommand(command, message) {
+    const filePath = process.env[`GITHUB_${command}`];
+    if (!filePath) {
+        throw new Error(`Unable to find environment variable for file command ${command}`);
+    }
+    if (!fs.existsSync(filePath)) {
+        throw new Error(`Missing file at path: ${filePath}`);
+    }
+    fs.appendFileSync(filePath, `${utils_1.toCommandValue(message)}${os.EOL}`, {
+        encoding: 'utf8'
+    });
+}
+exports.issueCommand = issueCommand;
+//# sourceMappingURL=file-command.js.map
+
+/***/ }),
+
 /***/ 118:
 /***/ (function(module, __unusedexports, __webpack_require__) {
 
@@ -3507,9 +4513,13 @@ var __importStar = (this && this.__importStar) || function (mod) {
     result["default"] = mod;
     return result;
 };
+var __importDefault = (this && this.__importDefault) || function (mod) {
+    return (mod && mod.__esModule) ? mod : { "default": mod };
+};
 Object.defineProperty(exports, "__esModule", { value: true });
 const core = __importStar(__webpack_require__(470));
 const github_1 = __webpack_require__(469);
+const minimatch_1 = __importDefault(__webpack_require__(93));
 function run() {
     var _a, _b, _c, _d;
     return __awaiter(this, void 0, void 0, function* () {
@@ -3517,7 +4527,7 @@ function run() {
             // Create GitHub client with the API token.
             const client = new github_1.GitHub(core.getInput('token', { required: true }));
             const format = core.getInput('format', { required: true });
-            const filter = core.getInput('filter', { required: true }) || '*';
+            const filter = core.getMultilineInput('filter', { required: true }) || '*';
             // Ensure that the format parameter is set properly.
             if (format !== 'space-delimited' && format !== 'csv' && format !== 'json') {
                 core.setFailed(`Format must be one of 'string-delimited', 'csv', or 'json', got '${format}'.`);
@@ -3567,9 +4577,22 @@ function run() {
                 core.setFailed(`The GitHub API for comparing the base and head commits for this ${github_1.context.eventName} event returned ${response.status}, expected 200. ` +
                     "Please submit an issue on this action's GitHub repo.");
             }
-            const regex = new RegExp(`/${filter}\\b`, 'g');
-            // Get the changed files from the response payload.
-            const files = response.data.files.filter(file => file.filename.match(regex));
+            const files = response.data.files.filter(file => {
+                let match = false;
+                for (const item of filter) {
+                    const pattern = item;
+                    core.debug(`Test ${file.filename} against ${pattern}`);
+                    core.debug(`current match value: ${match}`);
+                    if (pattern.startsWith('!')) {
+                        match = match && minimatch_1.default(file.filename, pattern, { matchBase: true, dot: true });
+                    }
+                    else {
+                        match = match || minimatch_1.default(file.filename, pattern, { matchBase: true, dot: true });
+                    }
+                    core.debug(`match: ${match}`);
+                }
+                return match;
+            });
             const all = [], added = [], modified = [], removed = [], renamed = [], addedModified = [];
             for (const file of files) {
                 const filename = file.filename;
@@ -4054,6 +5077,73 @@ function register (state, name, method, options) {
 }
 
 
+/***/ }),
+
+/***/ 284:
+/***/ (function(module) {
+
+"use strict";
+
+module.exports = balanced;
+function balanced(a, b, str) {
+  if (a instanceof RegExp) a = maybeMatch(a, str);
+  if (b instanceof RegExp) b = maybeMatch(b, str);
+
+  var r = range(a, b, str);
+
+  return r && {
+    start: r[0],
+    end: r[1],
+    pre: str.slice(0, r[0]),
+    body: str.slice(r[0] + a.length, r[1]),
+    post: str.slice(r[1] + b.length)
+  };
+}
+
+function maybeMatch(reg, str) {
+  var m = str.match(reg);
+  return m ? m[0] : null;
+}
+
+balanced.range = range;
+function range(a, b, str) {
+  var begs, beg, left, right, result;
+  var ai = str.indexOf(a);
+  var bi = str.indexOf(b, ai + 1);
+  var i = ai;
+
+  if (ai >= 0 && bi > 0) {
+    begs = [];
+    left = str.length;
+
+    while (i >= 0 && !result) {
+      if (i == ai) {
+        begs.push(i);
+        ai = str.indexOf(a, i + 1);
+      } else if (begs.length == 1) {
+        result = [ begs.pop(), bi ];
+      } else {
+        beg = begs.pop();
+        if (beg < left) {
+          left = beg;
+          right = bi;
+        }
+
+        bi = str.indexOf(b, i + 1);
+      }
+
+      i = ai < bi && ai >= 0 ? ai : bi;
+    }
+
+    if (begs.length) {
+      result = [ left, right ];
+    }
+  }
+
+  return result;
+}
+
+
 /***/ }),
 
 /***/ 293:
@@ -4390,6 +5480,214 @@ exports.paginateRest = paginateRest;
 //# sourceMappingURL=index.js.map
 
 
+/***/ }),
+
+/***/ 306:
+/***/ (function(module, __unusedexports, __webpack_require__) {
+
+var concatMap = __webpack_require__(896);
+var balanced = __webpack_require__(284);
+
+module.exports = expandTop;
+
+var escSlash = '\0SLASH'+Math.random()+'\0';
+var escOpen = '\0OPEN'+Math.random()+'\0';
+var escClose = '\0CLOSE'+Math.random()+'\0';
+var escComma = '\0COMMA'+Math.random()+'\0';
+var escPeriod = '\0PERIOD'+Math.random()+'\0';
+
+function numeric(str) {
+  return parseInt(str, 10) == str
+    ? parseInt(str, 10)
+    : str.charCodeAt(0);
+}
+
+function escapeBraces(str) {
+  return str.split('\\\\').join(escSlash)
+            .split('\\{').join(escOpen)
+            .split('\\}').join(escClose)
+            .split('\\,').join(escComma)
+            .split('\\.').join(escPeriod);
+}
+
+function unescapeBraces(str) {
+  return str.split(escSlash).join('\\')
+            .split(escOpen).join('{')
+            .split(escClose).join('}')
+            .split(escComma).join(',')
+            .split(escPeriod).join('.');
+}
+
+
+// Basically just str.split(","), but handling cases
+// where we have nested braced sections, which should be
+// treated as individual members, like {a,{b,c},d}
+function parseCommaParts(str) {
+  if (!str)
+    return [''];
+
+  var parts = [];
+  var m = balanced('{', '}', str);
+
+  if (!m)
+    return str.split(',');
+
+  var pre = m.pre;
+  var body = m.body;
+  var post = m.post;
+  var p = pre.split(',');
+
+  p[p.length-1] += '{' + body + '}';
+  var postParts = parseCommaParts(post);
+  if (post.length) {
+    p[p.length-1] += postParts.shift();
+    p.push.apply(p, postParts);
+  }
+
+  parts.push.apply(parts, p);
+
+  return parts;
+}
+
+function expandTop(str) {
+  if (!str)
+    return [];
+
+  // I don't know why Bash 4.3 does this, but it does.
+  // Anything starting with {} will have the first two bytes preserved
+  // but *only* at the top level, so {},a}b will not expand to anything,
+  // but a{},b}c will be expanded to [a}c,abc].
+  // One could argue that this is a bug in Bash, but since the goal of
+  // this module is to match Bash's rules, we escape a leading {}
+  if (str.substr(0, 2) === '{}') {
+    str = '\\{\\}' + str.substr(2);
+  }
+
+  return expand(escapeBraces(str), true).map(unescapeBraces);
+}
+
+function identity(e) {
+  return e;
+}
+
+function embrace(str) {
+  return '{' + str + '}';
+}
+function isPadded(el) {
+  return /^-?0\d/.test(el);
+}
+
+function lte(i, y) {
+  return i <= y;
+}
+function gte(i, y) {
+  return i >= y;
+}
+
+function expand(str, isTop) {
+  var expansions = [];
+
+  var m = balanced('{', '}', str);
+  if (!m || /\$$/.test(m.pre)) return [str];
+
+  var isNumericSequence = /^-?\d+\.\.-?\d+(?:\.\.-?\d+)?$/.test(m.body);
+  var isAlphaSequence = /^[a-zA-Z]\.\.[a-zA-Z](?:\.\.-?\d+)?$/.test(m.body);
+  var isSequence = isNumericSequence || isAlphaSequence;
+  var isOptions = m.body.indexOf(',') >= 0;
+  if (!isSequence && !isOptions) {
+    // {a},b}
+    if (m.post.match(/,.*\}/)) {
+      str = m.pre + '{' + m.body + escClose + m.post;
+      return expand(str);
+    }
+    return [str];
+  }
+
+  var n;
+  if (isSequence) {
+    n = m.body.split(/\.\./);
+  } else {
+    n = parseCommaParts(m.body);
+    if (n.length === 1) {
+      // x{{a,b}}y ==> x{a}y x{b}y
+      n = expand(n[0], false).map(embrace);
+      if (n.length === 1) {
+        var post = m.post.length
+          ? expand(m.post, false)
+          : [''];
+        return post.map(function(p) {
+          return m.pre + n[0] + p;
+        });
+      }
+    }
+  }
+
+  // at this point, n is the parts, and we know it's not a comma set
+  // with a single entry.
+
+  // no need to expand pre, since it is guaranteed to be free of brace-sets
+  var pre = m.pre;
+  var post = m.post.length
+    ? expand(m.post, false)
+    : [''];
+
+  var N;
+
+  if (isSequence) {
+    var x = numeric(n[0]);
+    var y = numeric(n[1]);
+    var width = Math.max(n[0].length, n[1].length)
+    var incr = n.length == 3
+      ? Math.abs(numeric(n[2]))
+      : 1;
+    var test = lte;
+    var reverse = y < x;
+    if (reverse) {
+      incr *= -1;
+      test = gte;
+    }
+    var pad = n.some(isPadded);
+
+    N = [];
+
+    for (var i = x; test(i, y); i += incr) {
+      var c;
+      if (isAlphaSequence) {
+        c = String.fromCharCode(i);
+        if (c === '\\')
+          c = '';
+      } else {
+        c = String(i);
+        if (pad) {
+          var need = width - c.length;
+          if (need > 0) {
+            var z = new Array(need + 1).join('0');
+            if (i < 0)
+              c = '-' + z + c.slice(1);
+            else
+              c = z + c;
+          }
+        }
+      }
+      N.push(c);
+    }
+  } else {
+    N = concatMap(n, function(el) { return expand(el, false) });
+  }
+
+  for (var j = 0; j < N.length; j++) {
+    for (var k = 0; k < post.length; k++) {
+      var expansion = pre + N[j] + post[k];
+      if (!isTop || isSequence || expansion)
+        expansions.push(expansion);
+    }
+  }
+
+  return expansions;
+}
+
+
+
 /***/ }),
 
 /***/ 323:
@@ -5230,15 +6528,29 @@ function octokitValidate(octokit) {
 
 "use strict";
 
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    Object.defineProperty(o, k2, { enumerable: true, get: function() { return m[k]; } });
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
+    Object.defineProperty(o, "default", { enumerable: true, value: v });
+}) : function(o, v) {
+    o["default"] = v;
+});
 var __importStar = (this && this.__importStar) || function (mod) {
     if (mod && mod.__esModule) return mod;
     var result = {};
-    if (mod != null) for (var k in mod) if (Object.hasOwnProperty.call(mod, k)) result[k] = mod[k];
-    result["default"] = mod;
+    if (mod != null) for (var k in mod) if (k !== "default" && Object.hasOwnProperty.call(mod, k)) __createBinding(result, mod, k);
+    __setModuleDefault(result, mod);
     return result;
 };
 Object.defineProperty(exports, "__esModule", { value: true });
+exports.issue = exports.issueCommand = void 0;
 const os = __importStar(__webpack_require__(87));
+const utils_1 = __webpack_require__(82);
 /**
  * Commands
  *
@@ -5293,13 +6605,13 @@ class Command {
     }
 }
 function escapeData(s) {
-    return (s || '')
+    return utils_1.toCommandValue(s)
         .replace(/%/g, '%25')
         .replace(/\r/g, '%0D')
         .replace(/\n/g, '%0A');
 }
 function escapeProperty(s) {
-    return (s || '')
+    return utils_1.toCommandValue(s)
         .replace(/%/g, '%25')
         .replace(/\r/g, '%0D')
         .replace(/\n/g, '%0A')
@@ -7268,6 +8580,25 @@ exports.GitHub = GitHub;
 
 "use strict";
 
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    Object.defineProperty(o, k2, { enumerable: true, get: function() { return m[k]; } });
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
+    Object.defineProperty(o, "default", { enumerable: true, value: v });
+}) : function(o, v) {
+    o["default"] = v;
+});
+var __importStar = (this && this.__importStar) || function (mod) {
+    if (mod && mod.__esModule) return mod;
+    var result = {};
+    if (mod != null) for (var k in mod) if (k !== "default" && Object.hasOwnProperty.call(mod, k)) __createBinding(result, mod, k);
+    __setModuleDefault(result, mod);
+    return result;
+};
 var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, generator) {
     function adopt(value) { return value instanceof P ? value : new P(function (resolve) { resolve(value); }); }
     return new (P || (P = Promise))(function (resolve, reject) {
@@ -7277,15 +8608,11 @@ var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, ge
         step((generator = generator.apply(thisArg, _arguments || [])).next());
     });
 };
-var __importStar = (this && this.__importStar) || function (mod) {
-    if (mod && mod.__esModule) return mod;
-    var result = {};
-    if (mod != null) for (var k in mod) if (Object.hasOwnProperty.call(mod, k)) result[k] = mod[k];
-    result["default"] = mod;
-    return result;
-};
 Object.defineProperty(exports, "__esModule", { value: true });
+exports.getState = exports.saveState = exports.group = exports.endGroup = exports.startGroup = exports.info = exports.warning = exports.error = exports.debug = exports.isDebug = exports.setFailed = exports.setCommandEcho = exports.setOutput = exports.getBooleanInput = exports.getMultilineInput = exports.getInput = exports.addPath = exports.setSecret = exports.exportVariable = exports.ExitCode = void 0;
 const command_1 = __webpack_require__(431);
+const file_command_1 = __webpack_require__(102);
+const utils_1 = __webpack_require__(82);
 const os = __importStar(__webpack_require__(87));
 const path = __importStar(__webpack_require__(622));
 /**
@@ -7308,11 +8635,21 @@ var ExitCode;
 /**
  * Sets env variable for this action and future actions in the job
  * @param name the name of the variable to set
- * @param val the value of the variable
+ * @param val the value of the variable. Non-string values will be converted to a string via JSON.stringify
  */
+// eslint-disable-next-line @typescript-eslint/no-explicit-any
 function exportVariable(name, val) {
-    process.env[name] = val;
-    command_1.issueCommand('set-env', { name }, val);
+    const convertedVal = utils_1.toCommandValue(val);
+    process.env[name] = convertedVal;
+    const filePath = process.env['GITHUB_ENV'] || '';
+    if (filePath) {
+        const delimiter = '_GitHubActionsFileCommandDelimeter_';
+        const commandValue = `${name}<<${delimiter}${os.EOL}${convertedVal}${os.EOL}${delimiter}`;
+        file_command_1.issueCommand('ENV', commandValue);
+    }
+    else {
+        command_1.issueCommand('set-env', { name }, convertedVal);
+    }
 }
 exports.exportVariable = exportVariable;
 /**
@@ -7328,12 +8665,20 @@ exports.setSecret = setSecret;
  * @param inputPath
  */
 function addPath(inputPath) {
-    command_1.issueCommand('add-path', {}, inputPath);
+    const filePath = process.env['GITHUB_PATH'] || '';
+    if (filePath) {
+        file_command_1.issueCommand('PATH', inputPath);
+    }
+    else {
+        command_1.issueCommand('add-path', {}, inputPath);
+    }
     process.env['PATH'] = `${inputPath}${path.delimiter}${process.env['PATH']}`;
 }
 exports.addPath = addPath;
 /**
- * Gets the value of an input.  The value is also trimmed.
+ * Gets the value of an input.
+ * Unless trimWhitespace is set to false in InputOptions, the value is also trimmed.
+ * Returns an empty string if the value is not defined.
  *
  * @param     name     name of the input to get
  * @param     options  optional. See InputOptions.
@@ -7344,19 +8689,70 @@ function getInput(name, options) {
     if (options && options.required && !val) {
         throw new Error(`Input required and not supplied: ${name}`);
     }
+    if (options && options.trimWhitespace === false) {
+        return val;
+    }
     return val.trim();
 }
 exports.getInput = getInput;
+/**
+ * Gets the values of an multiline input.  Each value is also trimmed.
+ *
+ * @param     name     name of the input to get
+ * @param     options  optional. See InputOptions.
+ * @returns   string[]
+ *
+ */
+function getMultilineInput(name, options) {
+    const inputs = getInput(name, options)
+        .split('\n')
+        .filter(x => x !== '');
+    return inputs;
+}
+exports.getMultilineInput = getMultilineInput;
+/**
+ * Gets the input value of the boolean type in the YAML 1.2 "core schema" specification.
+ * Support boolean input list: `true | True | TRUE | false | False | FALSE` .
+ * The return value is also in boolean type.
+ * ref: https://yaml.org/spec/1.2/spec.html#id2804923
+ *
+ * @param     name     name of the input to get
+ * @param     options  optional. See InputOptions.
+ * @returns   boolean
+ */
+function getBooleanInput(name, options) {
+    const trueValue = ['true', 'True', 'TRUE'];
+    const falseValue = ['false', 'False', 'FALSE'];
+    const val = getInput(name, options);
+    if (trueValue.includes(val))
+        return true;
+    if (falseValue.includes(val))
+        return false;
+    throw new TypeError(`Input does not meet YAML 1.2 "Core Schema" specification: ${name}\n` +
+        `Support boolean input list: \`true | True | TRUE | false | False | FALSE\``);
+}
+exports.getBooleanInput = getBooleanInput;
 /**
  * Sets the value of an output.
  *
  * @param     name     name of the output to set
- * @param     value    value to store
+ * @param     value    value to store. Non-string values will be converted to a string via JSON.stringify
  */
+// eslint-disable-next-line @typescript-eslint/no-explicit-any
 function setOutput(name, value) {
+    process.stdout.write(os.EOL);
     command_1.issueCommand('set-output', { name }, value);
 }
 exports.setOutput = setOutput;
+/**
+ * Enables or disables the echoing of commands into stdout for the rest of the step.
+ * Echoing is disabled by default if ACTIONS_STEP_DEBUG is not set.
+ *
+ */
+function setCommandEcho(enabled) {
+    command_1.issue('echo', enabled ? 'on' : 'off');
+}
+exports.setCommandEcho = setCommandEcho;
 //-----------------------------------------------------------------------
 // Results
 //-----------------------------------------------------------------------
@@ -7390,18 +8786,18 @@ function debug(message) {
 exports.debug = debug;
 /**
  * Adds an error issue
- * @param message error issue message
+ * @param message error issue message. Errors will be converted to string via toString()
  */
 function error(message) {
-    command_1.issue('error', message);
+    command_1.issue('error', message instanceof Error ? message.toString() : message);
 }
 exports.error = error;
 /**
  * Adds an warning issue
- * @param message warning issue message
+ * @param message warning issue message. Errors will be converted to string via toString()
  */
 function warning(message) {
-    command_1.issue('warning', message);
+    command_1.issue('warning', message instanceof Error ? message.toString() : message);
 }
 exports.warning = warning;
 /**
@@ -7459,8 +8855,9 @@ exports.group = group;
  * Saves state for current action, the state can only be retrieved by this action's post job execution.
  *
  * @param     name     name of the state to store
- * @param     value    value to store
+ * @param     value    value to store. Non-string values will be converted to a string via JSON.stringify
  */
+// eslint-disable-next-line @typescript-eslint/no-explicit-any
 function saveState(name, value) {
     command_1.issueCommand('save-state', { name }, value);
 }
@@ -24776,6 +26173,26 @@ function set(object, path, value) {
 module.exports = set;
 
 
+/***/ }),
+
+/***/ 896:
+/***/ (function(module) {
+
+module.exports = function (xs, fn) {
+    var res = [];
+    for (var i = 0; i < xs.length; i++) {
+        var x = fn(xs[i], i);
+        if (isArray(x)) res.push.apply(res, x);
+        else res.push(x);
+    }
+    return res;
+};
+
+var isArray = Array.isArray || function (xs) {
+    return Object.prototype.toString.call(xs) === '[object Array]';
+};
+
+
 /***/ }),
 
 /***/ 898:
diff --git a/package.json b/package.json
index 1279abb4..e0ed0da2 100644
--- a/package.json
+++ b/package.json
@@ -1,7 +1,7 @@
 {
   "name": "get-changed-files",
   "description": "GitHub action that gets all changed files in a pull request or push.",
-  "version": "1.0.0",
+  "version": "2.0.0",
   "private": true,
   "author": "Jitterbit, Inc.",
   "license": "MIT",
@@ -39,12 +39,13 @@
     "all": "yarn clean && yarn build && yarn format && yarn lint && yarn package && yarn test"
   },
   "dependencies": {
-    "@actions/core": "^1.2.2",
+    "@actions/core": "^1.4.0",
     "@actions/github": "^2.1.0"
   },
   "devDependencies": {
     "@octokit/rest": "^16.40.2",
     "@types/jest": "^25.1.1",
+    "@types/minimatch": "3.0.4",
     "@types/node": "^14.17.0",
     "@typescript-eslint/parser": "^2.8.0",
     "@zeit/ncc": "^0.21.0",
@@ -55,6 +56,7 @@
     "jest": "^25.1.0",
     "jest-circus": "^25.1.0",
     "js-yaml": "^3.13.1",
+    "minimatch": "^3.0.4",
     "prettier": "^1.19.1",
     "rimraf": "^3.0.0",
     "ts-jest": "^25.1.0",
diff --git a/src/main.ts b/src/main.ts
index 6f755aad..51a4a97e 100644
--- a/src/main.ts
+++ b/src/main.ts
@@ -1,5 +1,6 @@
 import * as core from '@actions/core'
 import {context, GitHub} from '@actions/github'
+import minimatch from 'minimatch'
 
 type Format = 'space-delimited' | 'csv' | 'json'
 type FileStatus = 'added' | 'modified' | 'removed' | 'renamed'
@@ -9,7 +10,7 @@ async function run(): Promise<void> {
     // Create GitHub client with the API token.
     const client = new GitHub(core.getInput('token', {required: true}))
     const format = core.getInput('format', {required: true}) as Format
-    const filter = core.getInput('filter', {required: true}) || '*'
+    const filter = core.getMultilineInput('filter', {required: true}) || '*'
 
     // Ensure that the format parameter is set properly.
     if (format !== 'space-delimited' && format !== 'csv' && format !== 'json') {
@@ -76,9 +77,22 @@ async function run(): Promise<void> {
       )
     }
 
-    const regex = new RegExp(`/${filter}\\b`, 'g')
-    // Get the changed files from the response payload.
-    const files = response.data.files.filter(file => file.filename.match(regex))
+    const files = response.data.files.filter(file => {
+      let match = false
+      for (const item of filter) {
+        const pattern = item
+        core.debug(`Test ${file.filename} against ${pattern}`)
+        core.debug(`current match value: ${match}`)
+        if (pattern.startsWith('!')) {
+          match = match && minimatch(file.filename, pattern, {matchBase: true, dot: true})
+        } else {
+          match = match || minimatch(file.filename, pattern, {matchBase: true, dot: true})
+        }
+        core.debug(`match: ${match}`)
+      }
+      return match
+    })
+
     const all = [] as string[],
       added = [] as string[],
       modified = [] as string[],