diff --git a/continuous_reporting/benchmark_and_update.rb b/continuous_reporting/benchmark_and_update.rb
index 03a830231..f4e2ac985 100755
--- a/continuous_reporting/benchmark_and_update.rb
+++ b/continuous_reporting/benchmark_and_update.rb
@@ -36,10 +36,10 @@
 # tolerance needs to be significantly larger to avoid frequent false positives.
 MICRO_DROP_TOLERANCE = 0.20
 
-YM_PAGES_DIR = File.expand_path "#{__dir__}/../../yjit-metrics-pages"
-YM_RAW_DATA_DIR = "#{YM_PAGES_DIR}/raw_benchmark_data"
+BUILT_REPORTS_ROOT = YJITMetrics::ContinuousReporting::BUILT_REPORTS_ROOT
+RAW_BENCHMARK_ROOT = YJITMetrics::ContinuousReporting::RAW_BENCHMARK_ROOT
 
-YM_REPORT_DIR = File.expand_path "#{YM_PAGES_DIR}/_includes/reports/"
+YM_REPORT_DIR = File.expand_path "#{BUILT_REPORTS_ROOT}/_includes/reports/"
 if File.exist?(YM_REPORT_DIR)
     var_warmup_reports = Dir.glob(YM_REPORT_DIR + "/variable_warmup_*.warmup_settings.json").to_a
     if var_warmup_reports.empty?
@@ -204,7 +204,7 @@ def ghapi_post(api_uri, params, verb: :post)
 class BenchmarkDetails
     def initialize(timestamp)
         @timestamp = timestamp
-        benchmark_details_file = File.join(YM_PAGES_DIR, "_benchmarks", "bench_#{timestamp}.md")
+        benchmark_details_file = File.join(BUILT_REPORTS_ROOT, "_benchmarks", "bench_#{timestamp}.md")
         @data = YAML.load File.read(benchmark_details_file)
     end
 
@@ -215,7 +215,7 @@ def raw_data
     def yjit_test_result
         yjit_file = @data["test_results"]["prod_ruby_with_yjit"]
         raise("Cannot locate latest YJIT data file for timestamp #{@timestamp}") unless yjit_file
-        File.join YM_PAGES_DIR, yjit_file
+        File.join RAW_BENCHMARK_ROOT, yjit_file
     end
 
     def yjit_permalink
@@ -329,7 +329,7 @@ def timestr_from_ts(ts)
 
 # If benchmark results drop noticeably, file a Github issue
 def check_perf_tripwires
-    Dir.chdir(__dir__ + "/../../yjit-metrics-pages/_includes/reports") do
+    Dir.chdir(YM_REPORT_DIR) do
         # Grab only non-platform-specific tripwire files that do *not* have a platform name in them,
         # but instead end in a six-digit timestamp.
         tripwire_files = Dir["*.tripwires.json"].to_a.select {|f| f =~ /\d{6}\.tripwires\.json\Z/}.sort
diff --git a/continuous_reporting/create_json_params_file.rb b/continuous_reporting/create_json_params_file.rb
index 09d2bea6b..a7bad1d55 100644
--- a/continuous_reporting/create_json_params_file.rb
+++ b/continuous_reporting/create_json_params_file.rb
@@ -57,7 +57,7 @@
 yjit_metrics_repo = ""
 yjit_bench_name = "main"
 yjit_bench_repo = "https://github.com/Shopify/yjit-bench.git"
-benchmark_data_dir = File.join(YJIT_METRICS_DIR, "raw-benchmark-data/raw_benchmark_data")
+benchmark_data_dir = nil
 
 # TODO: try looking up the given yjit_metrics and/or yjit_bench and/or CRuby revisions in the local repos to see if they exist?
 
@@ -108,10 +108,13 @@
   end
 
   opts.on("-bd PATH", "--benchmark-data-dir PATH") do |dir|
+    raise "--benchmark-data-dir must specify a directory" if dir.to_s.empty?
     benchmark_data_dir = File.expand_path(dir)
   end
 end.parse!
 
+raise "--benchmark-data-dir is required!" unless benchmark_data_dir
+
 def sha_for_name_in_dir(name:, dir:, repo:, desc:)
   Dir.chdir(dir) do
     system("git remote remove current_repo") # Don't care if this succeeds or not
diff --git a/continuous_reporting/file_benchmark_data_into_raw.rb b/continuous_reporting/file_benchmark_data_into_raw.rb
index d9c7963de..e53e8d4dc 100755
--- a/continuous_reporting/file_benchmark_data_into_raw.rb
+++ b/continuous_reporting/file_benchmark_data_into_raw.rb
@@ -7,15 +7,9 @@
 
 require_relative "../lib/yjit-metrics"
 
-YJIT_METRICS_PAGES_DIR = File.expand_path File.join(__dir__, "../../yjit-metrics-pages/raw_benchmark_data")
-unless File.exist?(YJIT_METRICS_PAGES_DIR)
-    raise "This script expects to be cloned in a repo right next to a \"yjit-metrics-pages\" repo of the `pages` branch of yjit-metrics"
-end
-
-YJIT_RAW_DATA_REPO = File.expand_path File.join(__dir__, "../../raw-benchmark-data/raw_benchmark_data")
+YJIT_RAW_DATA_REPO = File.join(YJITMetrics::ContinuousReporting::RAW_BENCHMARK_ROOT, "raw_benchmark_data")
 
 DESTINATIONS = [
-    #YJIT_METRICS_PAGES_DIR,
     YJIT_RAW_DATA_REPO,
 ]
 
diff --git a/continuous_reporting/generate_and_upload_reports.rb b/continuous_reporting/generate_and_upload_reports.rb
index 6b2af087b..fdf0e9969 100755
--- a/continuous_reporting/generate_and_upload_reports.rb
+++ b/continuous_reporting/generate_and_upload_reports.rb
@@ -9,25 +9,13 @@
 
 ### Required repos, etc, to build from
 
-# Dir in which yjit-metrics, yjit-bench, etc are cloned
-YM_ROOT_DIR = File.expand_path(File.join(__dir__, "../.."))
+YM_REPO = YJITMetrics::ContinuousReporting::YM_REPO
+RAW_BENCHMARK_ROOT = YJITMetrics::ContinuousReporting::RAW_BENCHMARK_ROOT
+RAW_REPORTS_ROOT = YJITMetrics::ContinuousReporting::RAW_REPORTS_ROOT
+BUILT_REPORTS_ROOT = YJITMetrics::ContinuousReporting::BUILT_REPORTS_ROOT
+GHPAGES_REPO = YJITMetrics::ContinuousReporting::GHPAGES_REPO
 
-# Clone of yjit-metrics repo, pages branch
-YJIT_METRICS_PAGES_DIR = File.expand_path File.join(YM_ROOT_DIR, "yjit-metrics-pages")
-
-# Raw benchmark data gets written to a platform- and date-specific subdirectory, but will often be read from multiple subdirectories
-RAW_BENCHMARK_ROOT = File.join(YM_ROOT_DIR, "raw-benchmark-data")
-
-# This contains Jekyll source files of various kinds - everything but the built reports
-RAW_REPORTS_ROOT = File.join(YM_ROOT_DIR, "raw-yjit-reports")
-
-# We cache all the built per-run reports, which can take a long time to rebuild
-BUILT_REPORTS_ROOT = File.join(YM_ROOT_DIR, "built-yjit-reports")
-
-# We have a separate repo for the final HTML, because creating the new orphan branch is fiddly
-GHPAGES_REPO = File.join(YM_ROOT_DIR, "ghpages-yjit-metrics")
-
-[YJIT_METRICS_PAGES_DIR, RAW_BENCHMARK_ROOT, RAW_REPORTS_ROOT, BUILT_REPORTS_ROOT].each do |dir|
+[RAW_BENCHMARK_ROOT, RAW_REPORTS_ROOT, BUILT_REPORTS_ROOT].each do |dir|
   unless File.exist?(dir)
     raise "We expected directory #{dir.inspect} to exist in order to generate reports!"
   end
@@ -137,10 +125,9 @@ def basic_report_filenames(report_name, ts, prefix: "#{BUILT_REPORTS_ROOT}/_incl
     REPORTS_AND_FILES.select! { |k, _| only_reports.include?(k) }
 end
 
-# From here on out, we're just in the yjit-metrics checkout of "pages" -- until we can stop relying on it.
-Dir.chdir(YJIT_METRICS_PAGES_DIR)
+# This is probably unnecessary now.
+Dir.chdir(YM_REPO)
 puts "Switched to #{Dir.pwd}"
-YJITMetrics.check_call("git checkout pages")
 
 # Turn JSON files into reports where outdated - first, find out what test results we have.
 # json_timestamps maps timestamps to file paths relative to the RAW_BENCHMARK_ROOT
@@ -214,7 +201,7 @@ def basic_report_filenames(report_name, ts, prefix: "#{BUILT_REPORTS_ROOT}/_incl
             reason ||= "we're missing files: #{missing_files.inspect}"
 
             puts "Running basic_report for timestamp #{ts} because #{reason} with data files #{test_files.inspect}"
-            YJITMetrics.check_call("ruby ../yjit-metrics/basic_report.rb -d #{RAW_BENCHMARK_ROOT} --report=#{report_name} -o #{BUILT_REPORTS_ROOT}/_includes/reports -w #{test_files.join(" ")}")
+            YJITMetrics.check_call("ruby #{YM_REPO}/basic_report.rb -d #{RAW_BENCHMARK_ROOT} --report=#{report_name} -o #{BUILT_REPORTS_ROOT}/_includes/reports -w #{test_files.join(" ")}")
 
             rf = basic_report_filenames(report_name, ts)
             files_not_found = rf.select { |f| !File.exist? f }
@@ -252,7 +239,7 @@ def basic_report_filenames(report_name, ts, prefix: "#{BUILT_REPORTS_ROOT}/_incl
                 # Add a field like blog_speed_details_x86_64_svg
                 YJITMetrics::PLATFORMS.each do |platform|
                     report_filename = "reports/#{report_name}_#{ts}.#{platform}.#{ext}"
-                    if File.exist?("_includes/#{report_filename}")
+                    if File.exist?("#{BUILT_REPORTS_ROOT}/_includes/#{report_filename}")
                         platforms[platform] = true
                         generated_reports[report_name + "_" + platform + "_" + ext.gsub(".", "_")] = report_filename
                     end
@@ -292,7 +279,7 @@ def basic_report_filenames(report_name, ts, prefix: "#{BUILT_REPORTS_ROOT}/_incl
 
     # It's possible to run only specific non-timeline reports -- then this would be empty.
     unless timeline_reports.empty?
-        YJITMetrics.check_call("ruby ../yjit-metrics/timeline_report.rb -d #{RAW_BENCHMARK_ROOT} --report='#{timeline_reports.keys.join(",")}' -o #{BUILT_REPORTS_ROOT}")
+        YJITMetrics.check_call("ruby #{YM_REPO}/timeline_report.rb -d #{RAW_BENCHMARK_ROOT} --report='#{timeline_reports.keys.join(",")}' -o #{BUILT_REPORTS_ROOT}")
     end
 
     # TODO: figure out a new way to verify that appropriate files were written. With various subdirs, the old way won't cut it.
diff --git a/continuous_reporting/gh_tasks/check_reporting.sh b/continuous_reporting/gh_tasks/check_reporting.sh
index b71694336..e863314b5 100755
--- a/continuous_reporting/gh_tasks/check_reporting.sh
+++ b/continuous_reporting/gh_tasks/check_reporting.sh
@@ -9,13 +9,10 @@ chruby 3.0.2
 cd ~/ym/yjit-metrics
 git pull
 
-# Copy benchmark raw data into yjit-metrics-pages repo
+# Copy benchmark raw data into destination repo
 #ruby continuous_reporting/file_benchmark_data_into_raw.rb -d continuous_reporting/data
 
-# Copy benchmark raw data into yjit-metrics-pages repo, generate reports, commit changes to Git.
+# Copy benchmark raw data into destination repo, generate reports, commit changes to Git.
 ruby continuous_reporting/generate_and_upload_reports.rb --no-push
 
-# Don't keep the provisional report files around
-cd ../yjit-metrics-pages && git checkout reports _includes
-
 echo "Reporting check completed successfully."
diff --git a/continuous_reporting/gh_tasks/check_reporting_minimal.sh b/continuous_reporting/gh_tasks/check_reporting_minimal.sh
index 77177b802..72248b751 100755
--- a/continuous_reporting/gh_tasks/check_reporting_minimal.sh
+++ b/continuous_reporting/gh_tasks/check_reporting_minimal.sh
@@ -11,13 +11,10 @@ chruby 3.0.2
 
 cd ~/ym/yjit-metrics
 
-# Copy benchmark raw data into yjit-metrics-pages repo
+# Copy benchmark raw data into destination repo
 #ruby continuous_reporting/file_benchmark_data_into_raw.rb -d continuous_reporting/data
 
 # Generate reports, commit changes to Git.
 ruby continuous_reporting/generate_and_upload_reports.rb --no-push --only-reports blog_timeline
 
-# Don't keep the provisional report files around
-cd ../yjit-metrics-pages && git checkout reports _includes && git clean -d -f reports _includes
-
 echo "Minimal reporting check completed successfully."
diff --git a/continuous_reporting/gh_tasks/report_and_upload.sh b/continuous_reporting/gh_tasks/report_and_upload.sh
index 656820343..8960d1503 100755
--- a/continuous_reporting/gh_tasks/report_and_upload.sh
+++ b/continuous_reporting/gh_tasks/report_and_upload.sh
@@ -10,7 +10,7 @@ chruby 3.0.2
 
 cd ~/ym/yjit-metrics
 
-# Copy benchmark raw data into yjit-metrics-pages repo, generate reports, commit changes to Git.
+# Copy benchmark raw data into destination repo, generate reports, commit changes to Git.
 ruby continuous_reporting/generate_and_upload_reports.rb
 
 # Now we'll verify that we're not regenerating results when we shouldn't.
diff --git a/continuous_reporting/gh_tasks/report_full_rebuild.sh b/continuous_reporting/gh_tasks/report_full_rebuild.sh
index 1f0dba279..b06d2bfae 100755
--- a/continuous_reporting/gh_tasks/report_full_rebuild.sh
+++ b/continuous_reporting/gh_tasks/report_full_rebuild.sh
@@ -10,7 +10,7 @@ chruby 3.0.2
 
 cd ~/ym/yjit-metrics
 
-# Copy benchmark raw data into yjit-metrics-pages repo, generate reports, commit changes to Git.
+# Copy benchmark raw data into destination repo, generate reports, commit changes to Git.
 # The --regenerate-reports argument will regenerate ***all*** reports, which can take quite a
 # long time. It will also occasionally hit a Ruby error, so we should update from 3.0.2 when
 # we can for "system" Ruby here.
diff --git a/continuous_reporting/jenkins/Jenkinsfile_data_checkin_and_upload b/continuous_reporting/jenkins/Jenkinsfile_data_checkin_and_upload
index 9f9b0d04b..171d3bb2a 100644
--- a/continuous_reporting/jenkins/Jenkinsfile_data_checkin_and_upload
+++ b/continuous_reporting/jenkins/Jenkinsfile_data_checkin_and_upload
@@ -19,7 +19,7 @@ pipeline {
     }
 
     stages {
-        // This stage commits and pushes the benchmark data from the yjit-metrics-pages directory, but doesn't get it there from yjit-metrics/continuous_reporting/data
+        // This stage commits and pushes the benchmark data from its directory, but doesn't get it there from yjit-metrics/continuous_reporting/data
         stage('data checkin') {
             parallel {
                 stage('data checkin x86_64') {
diff --git a/lib/yjit-metrics.rb b/lib/yjit-metrics.rb
index 461d5bf79..4fac2e215 100644
--- a/lib/yjit-metrics.rb
+++ b/lib/yjit-metrics.rb
@@ -9,6 +9,7 @@
 require_relative "./yjit-metrics/bench-results"
 require_relative "./yjit-metrics/repo-management"
 require_relative "./yjit-metrics/cli-interface"
+require_relative "./yjit-metrics/continuous_reporting"
 
 # Require all source files in yjit-metrics/report_types/*.rb
 Dir.glob("yjit-metrics/report_types/*.rb", base: __dir__).each do |report_type_file|
diff --git a/lib/yjit-metrics/continuous_reporting.rb b/lib/yjit-metrics/continuous_reporting.rb
new file mode 100644
index 000000000..b4835673e
--- /dev/null
+++ b/lib/yjit-metrics/continuous_reporting.rb
@@ -0,0 +1,21 @@
+module YJITMetrics
+  module ContinuousReporting
+    # Dir in which yjit-metrics, yjit-bench, etc are cloned
+    YM_ROOT_DIR = File.expand_path(File.join(__dir__, "../../.."))
+
+    # This repo.
+    YM_REPO = File.join(YM_ROOT_DIR, "yjit-metrics")
+
+    # Raw benchmark data gets written to a platform- and date-specific subdirectory, but will often be read from multiple subdirectories
+    RAW_BENCHMARK_ROOT = File.join(YM_ROOT_DIR, "raw-benchmark-data")
+
+    # This contains Jekyll source files of various kinds - everything but the built reports
+    RAW_REPORTS_ROOT = File.join(YM_ROOT_DIR, "raw-yjit-reports")
+
+    # We cache all the built per-run reports, which can take a long time to rebuild
+    BUILT_REPORTS_ROOT = File.join(YM_ROOT_DIR, "built-yjit-reports")
+
+    # We have a separate repo for the final HTML, because creating the new orphan branch is fiddly
+    GHPAGES_REPO = File.join(YM_ROOT_DIR, "ghpages-yjit-metrics")
+  end
+end