diff --git a/.github/copilot-instructions.md b/.github/copilot-instructions.md
new file mode 100644
index 00000000..daad4f0d
--- /dev/null
+++ b/.github/copilot-instructions.md
@@ -0,0 +1,28 @@
+# Copilot Instructions for pretext-cli
+
+Prefer running project commands with `poetry run` so they use the repository's
+configured Python environment and tool versions.
+
+When working on an assigned GitHub issue and preparing to open a pull request,
+run code formatting before creating the PR:
+
+```bash
+poetry run black .
+```
+
+Only open the PR after formatting has been run successfully.
+
+When changes affect behavior in `pretext/`, add or update tests in `tests/`
+that cover the change.
+
+Prefer test-driven development whenever practical: write or update tests first,
+then implement the code change to satisfy them.
+
+Before opening the PR, run relevant tests (or the full suite when needed):
+
+```bash
+poetry run pytest
+```
+
+For user-visible changes, add an entry under `[Unreleased]` in `CHANGELOG.md`
+using Keep a Changelog categories (Added, Changed, Fixed, Removed).
diff --git a/pretext/project/__init__.py b/pretext/project/__init__.py
index 73838c40..43f40a37 100644
--- a/pretext/project/__init__.py
+++ b/pretext/project/__init__.py
@@ -61,6 +61,16 @@ class Format(str, Enum):
     CUSTOM = "custom"
 
 
+# Maps single-file output formats to their file extension, used when looking up
+# the actual output file to generate a better deploy link.
+_SINGLE_FILE_FORMAT_EXTENSIONS: t.Dict["Format", str] = {
+    Format.PDF: ".pdf",
+    Format.EPUB: ".epub",
+    Format.KINDLE: ".epub",
+    Format.BRAILLE: ".brl",
+}
+
+
 # The CLI only needs two values from the publication file. Therefore, this class ignores the vast majority of a publication file's contents, loading and validating only a (small) relevant subset.
 # Since we will want to hash the baseurl for generating qr codes, we also load it here.
 class PublicationSubset(
@@ -439,9 +449,19 @@ def deploy_dir_relpath(self) -> Path:
         return self._project.stage / self.deploy_dir_path()
 
     def deploy_path(self) -> Path:
-        if self.output_filename is None:
-            return self.deploy_dir_path()
-        return self.deploy_dir_path() / self.output_filename
+        if self.output_filename is not None:
+            return self.deploy_dir_path() / self.output_filename
+        # For single-file output formats, look for the actual output file in
+        # the output directory to create a better link in the pelican-generated site.
+        ext = _SINGLE_FILE_FORMAT_EXTENSIONS.get(self.format)
+        if ext is not None:
+            output_dir = self.output_dir_abspath()
+            if output_dir.exists():
+                gen = output_dir.glob(f"*{ext}")
+                first = next(gen, None)
+                if first is not None and next(gen, None) is None:
+                    return self.deploy_dir_path() / first.name
+        return self.deploy_dir_path()
 
     def xsl_abspath(self) -> t.Optional[Path]:
         if self.xsl is None:
diff --git a/tests/test_project.py b/tests/test_project.py
index 19dddb83..9e591da1 100644
--- a/tests/test_project.py
+++ b/tests/test_project.py
@@ -416,6 +416,52 @@ def test_deploy(tmp_path: Path) -> None:
         )
 
 
+def test_deploy_path(tmp_path: Path) -> None:
+    # Test that deploy_path() returns the correct path for single-file formats
+    # when output_filename is not specified.
+    prj_path = tmp_path / "test_deploy_path"
+    shutil.copytree(EXAMPLES_DIR / "projects" / "project_refactor" / "simple", prj_path)
+    (prj_path / "project.ptx").unlink()
+    with utils.working_directory(prj_path):
+        project = pr.Project(ptx_version="2")
+
+        # For PDF target with no output_filename and no output dir: fallback to directory
+        t_pdf = project.new_target(name="print", format="pdf", deploy_dir="print-dir")
+        assert t_pdf.deploy_path() == Path("print-dir")
+
+        # Simulate a built PDF in the output directory
+        pdf_output_dir = t_pdf.output_dir_abspath()
+        pdf_output_dir.mkdir(parents=True, exist_ok=True)
+        (pdf_output_dir / "main.pdf").touch()
+
+        # deploy_path() should now find the PDF and include it in the path
+        assert t_pdf.deploy_path() == Path("print-dir") / "main.pdf"
+
+        # If output_filename is explicitly set, it should take precedence
+        t_pdf_explicit = project.new_target(
+            name="print-explicit",
+            format="pdf",
+            deploy_dir="print-dir2",
+            output_filename="custom.pdf",
+        )
+        assert t_pdf_explicit.deploy_path() == Path("print-dir2") / "custom.pdf"
+
+        # For HTML format (directory output), deploy_path() should still return the directory
+        t_html = project.new_target(name="web", format="html", deploy_dir="web-dir")
+        assert t_html.deploy_path() == Path("web-dir")
+
+        # EPUB target with no output file: fallback to directory
+        t_epub = project.new_target(name="epub", format="epub", deploy_dir="epub-dir")
+        assert t_epub.deploy_path() == Path("epub-dir")
+
+        # Simulate a built EPUB
+        epub_output_dir = t_epub.output_dir_abspath()
+        epub_output_dir.mkdir(parents=True, exist_ok=True)
+        (epub_output_dir / "book.epub").touch()
+
+        assert t_epub.deploy_path() == Path("epub-dir") / "book.epub"
+
+
 def test_validation(tmp_path: Path) -> None:
     project = pr.Project(ptx_version="2")
     # Verify that repeated server names cause a validation error.