diff --git a/.dockstore.yml b/.dockstore.yml
new file mode 100644
index 000000000..a6bc6cf36
--- /dev/null
+++ b/.dockstore.yml
@@ -0,0 +1,8 @@
+version: 1.2
+workflows:
+   - subclass: WDL
+     primaryDescriptorPath: /workflow/wdl/workflow.wdl
+     testParameterFiles:
+     - /workflow/wdl/input_ncov.json
+     authors:
+      - name: Nextstrain
\ No newline at end of file
diff --git a/workflow/wdl/input_ncov.json b/workflow/wdl/input_ncov.json
new file mode 100644
index 000000000..371113038
--- /dev/null
+++ b/workflow/wdl/input_ncov.json
@@ -0,0 +1,5 @@
+{
+  "Nextstrain_WRKFLW.build_yaml" : "data/build.yaml",
+  "Nextstrain_WRKFLW.giturl" : "https://github.com/nextstrain/ncov/archive/refs/heads/master.zip",
+  "Nextstrain_WRKFLW.docker_path": "nextstrain/base:latest"
+}
diff --git a/workflow/wdl/workflow.wdl b/workflow/wdl/workflow.wdl
new file mode 100644
index 000000000..2e02cd504
--- /dev/null
+++ b/workflow/wdl/workflow.wdl
@@ -0,0 +1,139 @@
+version 1.0
+
+# import "tasks/nextstrain.wdl" as nextstrain  # <= modular method
+
+workflow Nextstrain_WRKFLW {
+  input {
+    # Option 1: run the ncov example workflow
+    File? build_yaml
+
+    # Option 2: create a build_yaml from sequence and metadata
+    File? sequence_fasta
+    File? metadata_tsv
+    String? build_name
+    # It's possible all of the above files are provided
+
+    # Option 3? GISAID augur zip?
+    # File? gisaid_zip # tarball
+
+    String? active_builds # "Wisconsin,Minnesota,Iowa"
+
+    # By default, run the ncov workflow (can swap it for zika or something else)
+    String giturl = "https://github.com/nextstrain/ncov/archive/refs/heads/master.zip"
+    String docker_path = "nextstrain/base:latest"
+    Int? cpu
+    Int? memory       # in GiB
+    Int? disk_size
+  }
+
+  if (defined(sequence_fasta)) {
+    call mk_buildconfig {
+      input:
+        sequence_fasta = select_first([sequence_fasta]),
+        metadata_tsv = select_first([metadata_tsv]),
+        build = build_name,
+        dockerImage = docker_path
+    }
+  }
+
+  # call nextstrain.nextstrain build as build {  # <= modular method
+  call nextstrain_build as build {
+    input:
+      build_yaml = select_first([build_yaml, mk_buildconfig.buildconfig]), # Accepts Option 1 or Option 2
+      cpu = cpu,
+      memory = memory,
+      disk_size = disk_size,
+      dockerImage = docker_path,
+      giturl = giturl,
+      active_builds = active_builds
+  }
+
+  output {
+    Array[File] json_files = build.json_files
+    File auspice_zip = build.auspice_zip
+  }
+}
+
+# === Define Tasks
+task mk_buildconfig {
+  input {
+    File sequence_fasta  # Could change this to Array[File] and loop the HEREDOC
+    File metadata_tsv
+    String build = "example"
+    String dockerImage
+    Int cpu = 1
+    Int disk_size = 5
+    Float memory = 3.5
+  }
+  command {
+    cat << EOF > build.yaml
+    inputs:
+    - name: ~{build}
+      metadata: ~{metadata_tsv}
+      sequences: ~{sequence_fasta}
+    - name: references
+      metadata: data/references_metadata.tsv
+      sequences: data/references_sequences.fasta
+    EOF
+  }
+  output {
+    File buildconfig = "build.yaml"
+  }
+  runtime {
+    docker: dockerImage
+    cpu : cpu
+    memory: memory + " GiB"
+    disks: "local-disk " + disk_size + " HDD"
+  }
+}
+
+# Public Reference Datasets in case we want to add default "context" strains for ncov
+# nextstrain remote ls s3://nextstrain-data/files/ncov/open &> list.txt
+# files/ncov/open/LOCATION/metadata.tsv.xz where LOCATION = one of ['africa', 'asia', 'europe', 'global', 'north-america', 'oceania', 'south-america']
+# files/ncov/open/LOCATION/sequences.fasta.xz
+# TODO: since these are all s3, just generate string (avoid download and localazation/delocalization) for the mk_buildconfig
+# Clarification: Do these need to be sampled down?
+
+task nextstrain_build {
+  input {
+    File? build_yaml
+    String? active_builds # Wisconsin,Minnesota,Washington
+    String dockerImage = "nextstrain/base:latest"
+    String nextstrain_app = "nextstrain"
+    String giturl = "https://github.com/nextstrain/ncov/archive/refs/heads/master.zip"
+    Int cpu = 8         # Honestly, I'd max this out unless budget is a consideration.
+    Int disk_size = 30  # In GiB.  Could also check size of sequence or metadata files
+    Float memory = 3.5 
+  }
+  command {
+    # Pull ncov, zika or similar repository
+    wget -O master.zip ~{giturl}
+    INDIR=`unzip -Z1 master.zip | head -n1 | sed 's:/::g'`
+    unzip master.zip  
+    
+    # Max out the number of threads
+    PROC=`nproc`  
+
+    # Run nextstrain
+    "~{nextstrain_app}" build \
+      --cpus $PROC \
+      --memory  ~{memory}Gib \
+      --native $INDIR ~{"--configfile " + build_yaml} \
+      ~{"--config active_builds=" + active_builds}
+      
+    # Prepare output
+    mv $INDIR/auspice .
+    zip -r auspice.zip auspice
+  }
+  output {
+    File auspice_zip = "auspice.zip"
+    Array[File] json_files = glob("auspice/*.json")
+    # Target the s3
+  }
+  runtime {
+    docker: dockerImage
+    cpu : cpu
+    memory: memory + " GiB"
+    disks: "local-disk " + disk_size + " HDD"
+  }
+}