diff --git a/download_action.h b/download_action.h
new file mode 100644
index 0000000..9abbdcf
--- /dev/null
+++ b/download_action.h
@@ -0,0 +1,110 @@
+// Copyright (c) 2009 The Chromium OS Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef UPDATE_ENGINE_DOWNLOAD_ACTION_H__
+#define UPDATE_ENGINE_DOWNLOAD_ACTION_H__
+
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+
+#include <map>
+#include <string>
+
+#include <curl/curl.h>
+
+#include "base/scoped_ptr.h"
+#include "update_engine/action.h"
+#include "update_engine/decompressing_file_writer.h"
+#include "update_engine/file_writer.h"
+#include "update_engine/http_fetcher.h"
+#include "update_engine/omaha_hash_calculator.h"
+
+// The Download Action downloads a requested url to a specified path on disk.
+// It takes no input object, but if successful, passes the output path
+// to the output pipe.
+
+using std::map;
+using std::string;
+
+namespace chromeos_update_engine {
+
+class DownloadAction;
+class NoneType;
+
+template<>
+class ActionTraits<DownloadAction> {
+ public:
+  // Does not take an object for input
+  typedef NoneType InputObjectType;
+  // On success, puts the output path on output
+  typedef std::string OutputObjectType;
+};
+
+class DownloadAction : public Action<DownloadAction>,
+                       public HttpFetcherDelegate {
+ public:
+  // Takes ownership of the passed in HttpFetcher. Useful for testing.
+  // A good calling pattern is:
+  // DownloadAction(..., new WhateverHttpFetcher);
+  DownloadAction(const std::string& url, const std::string& output_path,
+                 off_t size, const std::string& hash,
+                 const bool should_decompress,
+                 HttpFetcher* http_fetcher);
+  virtual ~DownloadAction();
+  typedef ActionTraits<DownloadAction>::InputObjectType InputObjectType;
+  typedef ActionTraits<DownloadAction>::OutputObjectType OutputObjectType;
+  void PerformAction();
+  void TerminateProcessing();
+
+  // Debugging/logging
+  std::string Type() const { return "DownloadAction"; }
+
+  // Delegate methods (see http_fetcher.h)
+  virtual void ReceivedBytes(HttpFetcher *fetcher,
+                             const char* bytes, int length);
+  virtual void TransferComplete(HttpFetcher *fetcher, bool successful);
+
+ private:
+  // Expected size of the file (will be used for progress info)
+  const size_t size_;
+
+  // URL to download
+  const std::string url_;
+
+  // Path to save URL to
+  const std::string output_path_;
+
+  // Expected hash of the file. The hash must match for this action to
+  // succeed.
+  const std::string hash_;
+
+  // Whether the caller requested that we decompress the downloaded data.
+  const bool should_decompress_;
+
+  // The FileWriter that downloaded data should be written to. It will
+  // either point to *decompressing_file_writer_ or *direct_file_writer_.
+  FileWriter* writer_;
+
+  // If non-null, a FileWriter used for gzip decompressing downloaded data
+  scoped_ptr<GzipDecompressingFileWriter> decompressing_file_writer_;
+
+  // Used to write out the downloaded file
+  scoped_ptr<DirectFileWriter> direct_file_writer_;
+
+  // pointer to the HttpFetcher that does the http work
+  scoped_ptr<HttpFetcher> http_fetcher_;
+
+  // Used to find the hash of the bytes downloaded
+  OmahaHashCalculator omaha_hash_calculator_;
+  DISALLOW_COPY_AND_ASSIGN(DownloadAction);
+};
+
+// We want to be sure that we're compiled with large file support on linux,
+// just in case we find ourselves downloading large images.
+COMPILE_ASSERT(8 == sizeof(off_t), off_t_not_64_bit);
+
+}  // namespace chromeos_update_engine
+
+#endif  // UPDATE_ENGINE_DOWNLOAD_ACTION_H__