@@ -22,6 +22,10 @@ inputs:
     description: Actual number of synchronized branches.
     type: string
     required: true
+  created-tags:
+    description: Actual number of created tags.
+    type: string
+    required: true
   skipped:
     description: Actual number of skipped entries.
     type: string
@@ -34,6 +38,10 @@ inputs:
     description: Expected number of synchronized branches.
     type: string
     required: true
+  expected-created-tags:
+    description: Expected number of created tags.
+    type: string
+    required: true
   expected-skipped:
     description: Expected number of skipped entries.
     type: string
@@ -50,9 +58,11 @@ runs:
       shell: bash
       env:
         SYNCHRONIZED:          ${{ inputs.synchronized }}
+        CREATED_TAGS:          ${{ inputs.created-tags }}
         SKIPPED:               ${{ inputs.skipped }}
         ERRORS:                ${{ inputs.errors }}
         EXPECTED_SYNCHRONIZED: ${{ inputs.expected-synchronized }}
+        EXPECTED_CREATED_TAGS: ${{ inputs.expected-created-tags }}
         EXPECTED_SKIPPED:      ${{ inputs.expected-skipped }}
         EXPECTED_ERRORS:       ${{ inputs.expected-errors }}
       run: |
@@ -78,6 +88,7 @@ runs:
         }
 
         checkOutput "synchronized" "${SYNCHRONIZED}" "${EXPECTED_SYNCHRONIZED}"
+        checkOutput "created-tags" "${CREATED_TAGS}" "${EXPECTED_CREATED_TAGS}"
         checkOutput "skipped"      "${SKIPPED}"      "${EXPECTED_SKIPPED}"
         checkOutput "errors"       "${ERRORS}"       "${EXPECTED_ERRORS}"
 
 
@@ -43,11 +43,14 @@ jobs:
         uses: ./.github/actions/check-outputs
         with:
           synchronized:          ${{ steps.synchronize.outputs.synchronized }}
+          created-tags:          ${{ steps.synchronize.outputs.created-tags }}
           skipped:               ${{ steps.synchronize.outputs.skipped }}
           errors:                ${{ steps.synchronize.outputs.errors }}
           # 1 branch of 'ghdl' + 1 branch of 'progit2' + 2 branches of 'itertree'
           expected-synchronized: 4
           # 1 commented out organisation + 1 commented out repository
+          # Dry-run reads no repository, so no tag is created.
+          expected-created-tags: 0
           expected-skipped:      2
           expected-errors:       0
 
@@ -73,9 +76,11 @@ jobs:
         uses: ./.github/actions/check-outputs
         with:
           synchronized:          ${{ steps.synchronize.outputs.synchronized }}
+          created-tags:          ${{ steps.synchronize.outputs.created-tags }}
           skipped:               ${{ steps.synchronize.outputs.skipped }}
           errors:                ${{ steps.synchronize.outputs.errors }}
           expected-synchronized: 1
+          expected-created-tags: 0
           expected-skipped:      0
           # 1 missing 'Missing.repos' + 4 malformed lines in 'ghdl.repos'
           expected-errors:       5
 
@@ -3,8 +3,8 @@
 
 # Synchronize Forks
 
-A reusable GitHub Action synchronizing forked repositories of a GitHub organisation or user account with their upstream
-repositories.
+A reusable GitHub Action synchronizing the branches and tags of forked repositories of a GitHub organisation or user
+account with their upstream repositories.
 
 Forks don't update themselves. GitHub offers a *Sync fork* button per repository and per branch, but no automation for
 a whole namespace. This action reads a list of forks and their branches from simple configuration files and
@@ -86,27 +86,28 @@ file each, so they are collected in an `_Others` (or `_Misc`) file.
 
 Each of these files lists one fork per line, in the format:
 
-`<upstream>=<fork>:<branches>`
+`<upstream>=<fork>:<branches>[:<tagPatterns>]`
 
-| Element      | Format                                            | Meaning                                                                                                                                 |
-|--------------|---------------------------------------------------|-----------------------------------------------------------------------------------------------------------------------------------------|
-| `<upstream>` | `<organisation>/<repository>`                     | The repository the fork was created from. It's reported in the action's progress and error output, so it should name the real upstream.  |
-| `<fork>`     | `<repository>`                                    | The forked repository. No organisation or account, because that's the `target-organisation` parameter.                                   |
-| `<branches>` | `<branch>[,<branch>,...]`                         | Comma separated list of branches to synchronize.                                                                                        |
+| Element         | Format                        | Meaning                                                                                                                                 |
+|-----------------|-------------------------------|-------------------------------------------------------------------------------------------------------------------------------------------|
+| `<upstream>`    | `<organisation>/<repository>` | The repository the fork was created from. It's reported in the action's progress and error output, so it should name the real upstream.  |
+| `<fork>`        | `<repository>`                | The forked repository. No organisation or account, because that's the `target-organisation` parameter.                                   |
+| `<branches>`    | `<branch>[,<branch>,...]`     | Comma separated list of branches to synchronize. May be empty when tag patterns are given.                                               |
+| `<tagPatterns>` | `<pattern>[,<pattern>,...]`   | *Optional.* Comma separated list of tag names or regular expressions. See [Tag Synchronization](#tag-synchronization).                   |
 
 A line starting with `#` is a comment and skips that fork. Empty lines are ignored.
 
 ```
-OSVVM/OSVVM=OSVVM:main,dev
+OSVVM/OSVVM=OSVVM:main,dev:v\d+\.\d+.*
 OSVVM/OSVVM-Scripts=OSVVM-Scripts:main,dev
 OSVVM/AXI4=OSVVM-AXI4:main,dev
 OSVVM/Ethernet=OSVVM-Ethernet:main
 #OSVVM/AvalonST=OSVVM-AvalonST:main
 ```
 
-Read for the `PLC2` namespace, those five lines synchronize `PLC2/OSVVM` (branches `main` and `dev`) from
-`OSVVM/OSVVM`, `PLC2/OSVVM-Scripts` from `OSVVM/OSVVM-Scripts`, `PLC2/OSVVM-AXI4` from `OSVVM/AXI4`,
-`PLC2/OSVVM-Ethernet` (branch `main` only) from `OSVVM/Ethernet`, and skip `PLC2/OSVVM-AvalonST`.
+Read for the `PLC2` namespace, those five lines synchronize `PLC2/OSVVM` (branches `main` and `dev`, plus every tag
+matching `v\d+\.\d+.*`) from `OSVVM/OSVVM`, `PLC2/OSVVM-Scripts` from `OSVVM/OSVVM-Scripts`, `PLC2/OSVVM-AXI4` from
+`OSVVM/AXI4`, `PLC2/OSVVM-Ethernet` (branch `main` only) from `OSVVM/Ethernet`, and skip `PLC2/OSVVM-AvalonST`.
 
 Both Unix and Windows line endings are accepted, and a missing final newline doesn't drop the last entry.
 
@@ -119,13 +120,16 @@ Every organisation becomes a collapsible group, and each fork reports the branch
   📂 OSVVM/OSVVM ⇒ PLC2/OSVVM
     ✅ gh repo sync PLC2/OSVVM --branch main
     ✅ gh repo sync PLC2/OSVVM --branch dev
+    🏷️ v2.1.0 — created from OSVVM/OSVVM@a1b2c3d
+    🏷️ 12 tag(s) already up to date
   📂 OSVVM/AXI4 ⇒ PLC2/OSVVM-AXI4
     ❌ gh repo sync PLC2/OSVVM-AXI4 --branch dev
     ↪ failed to sync: HTTP 404: Not Found
   🚫 #OSVVM/AvalonST=OSVVM-AvalonST:main
 
 Summary:
   Synchronized branches:  2
+  Created tags:           1
   Skipped entries:        1
   Errors:                 1
 
@@ -150,10 +154,43 @@ Not synchronized:
 | Parameter      | Description                                                             |
 |----------------|---------------------------------------------------------------------------|
 | `synchronized` | Number of successfully synchronized branches.                           |
+| `created-tags` | Number of tags created in a fork.                                       |
 | `skipped`      | Number of skipped organisations and repositories (commented out lines). |
 | `errors`       | Number of counted errors.                                               |
 
-In dry-run mode, `synchronized` counts the branches that *would* have been synchronized.
+In dry-run mode, `synchronized` counts the branches that *would* have been synchronized, the configured tag patterns
+are printed, and no repository is read.
+
+## Tag Synchronization
+
+`gh repo sync` knows branches only — a fork's tags are never updated by it, which is why a fork drifts behind its
+upstream in releases even while its branches are current. `Paebbels/ghdl` is a live example: `ghdl/ghdl` has 46 tags,
+the fork has 19.
+
+The optional fourth field of a configuration line says which tags to follow:
+
+```
+ghdl/ghdl=ghdl:master:v\d+\.\d+.*
+OSVVM/OSVVM=OSVVM:main,dev:nightly,v\d+\.\d+\.\d+
+antonblanchard/microwatt=microwatt::v\d+\.\d+
+```
+
+A pattern is either a fixed tag name or a regular expression, and it has to match the **whole** tag name. Patterns are
+comma separated, and a line may carry tags without any branch, as the third line shows. Matching is done with
+`grep -P`, so PCRE syntax including `\d`, `\w` and `{n,m}` is available. Beware that a fixed name is a regular
+expression too: `v1.0` matches `v1x0` as well.
+
+For every matching tag of the upstream repository:
+
+* the fork doesn't have it → it's created, pointing at the same object:  
+  `🏷️ v2.1.0 — created from OSVVM/OSVVM@a1b2c3d`
+* the fork has it, at the same object → counted as up to date, reported as one line per repository,
+* the fork has it, at a **different** object → the tag moved upstream. It's reported as an error and **left alone**,
+  because rewriting it would silently discard whatever the fork's tag points at:  
+  `❌ v1.0.0 — moved in 'OSVVM/OSVVM' (fork 90e6af7, upstream d3d07ba)`  
+  The run continues with the next tag. Delete the tag in the fork to let the next run recreate it.
+
+Tags are never deleted from the fork, and a tag that exists only in the fork is left untouched.
 
 ## Error Handling
 
@@ -164,7 +201,9 @@ Each of these is counted, reported as a GitHub Actions error annotation, and let
 * an `<organisation>.repos` file listed in the index file doesn't exist,
 * a repository line is malformed — no `=`, no `:`, or an empty `<upstream>`, `<fork>` or branch list,
 * `gh repo sync` fails for a fork's branch. Its output is quoted below the failed command, and the annotation names the
-  upstream repository, the fork and the branch.
+  upstream repository, the fork and the branch,
+* a tag moved in the upstream repository, or a tag can't be created in the fork,
+* the tags of a repository can't be read, or the runner's `grep` has no `-P` support.
 
 A failing fork doesn't stop the run: every other fork is still synchronized, and the summary lists what wasn't.
 
@@ -185,12 +224,15 @@ A repository that carries the inline script keeps its `*.repos` files and replac
 with the `uses:` step shown above. `targetOrganisation=<name>`, the constant that had to be edited in every copy,
 becomes the `target-organisation` parameter — and can be dropped where the namespace is the one the workflow runs in.
 
-The configuration file format is unchanged. One field is worth a look while converting: `<upstream>` is reported in the
-progress and error output, so a copied placeholder there makes the log name the wrong repository.
+The configuration file format is backwards compatible — a line without the tag field behaves exactly as before. Two
+fields are worth a look while converting: `<upstream>` is reported in the progress and error output, so a copied
+placeholder there makes the log name the wrong repository, and adding tag patterns is what stops the fork from drifting
+behind in releases.
 
 ## Dependencies
 
 * [GitHub CLI (`gh`)](https://cli.github.com/), pre-installed on GitHub-hosted runners.
+* `grep` with PCRE support (`-P`), for tag patterns only. GNU grep on the Linux runners has it.
 
 ## Contributors