diff --git a/.github/workflows/ci.yaml b/.github/workflows/ci.yaml
index 7727ade..db35448 100644
--- a/.github/workflows/ci.yaml
+++ b/.github/workflows/ci.yaml
@@ -1,7 +1,6 @@
 name: ci
 
 on:
-  push:
   pull_request:
 
 env:
diff --git a/DESCRIPTION b/DESCRIPTION
index d06cbe0..1f7e403 100644
--- a/DESCRIPTION
+++ b/DESCRIPTION
@@ -1,9 +1,11 @@
 Package: compost
-Version: 0.1.0.7
-Title: Video Compositing via FFmpeg
-Description: Wraps common FFmpeg filter_complex patterns (overlay, chromakey,
-    concat, scale, vstack) into clean R functions for video compositing, and
-    lowers an OpenTimelineIO timeline (via rotio) to a rendered video.
+Version: 0.1.0.8
+Title: Video Compositing via 'FFmpeg'
+Description: Wraps common 'FFmpeg' <https://ffmpeg.org/> filter_complex
+    patterns (overlay, chromakey, concat, scale, vstack) into clean R
+    functions for video compositing, and lowers an 'OpenTimelineIO'
+    <https://opentimeline.io/> timeline (via the 'rotio' package) to a
+    rendered video.
 Authors@R: c(
     person("Troy", "Hernandez",
         email = "troy@cornball.ai",
@@ -11,6 +13,8 @@ Authors@R: c(
         comment = c(ORCID = "0009-0005-4248-604X")),
     person("cornball.ai", role = "cph"))
 License: MIT + file LICENSE
+URL: https://github.com/cornball-ai/compost
+BugReports: https://github.com/cornball-ai/compost/issues
 Encoding: UTF-8
 SystemRequirements: ffmpeg (>= 5.0)
 Imports:
diff --git a/R/align.R b/R/align.R
index 1006d83..9bc433e 100644
--- a/R/align.R
+++ b/R/align.R
@@ -107,6 +107,10 @@ align_audio <- function(clip, narration, sr = 8000L, hop = 80L) {
 #' @return A data frame with one row per chunk: \code{chunk}, \code{file},
 #'   \code{frames}, \code{fps}, \code{offset} (narration seconds),
 #'   \code{correlation}, and \code{overlap} (frames; 0 for the first chunk).
+#' @examples
+#' \dontrun{
+#' align_overlaps(c("media/chunk01.mp4", "media/chunk02.mp4"), "audio.mp3")
+#' }
 #' @export
 align_overlaps <- function(files, narration, fps = 24) {
     n <- length(files)
diff --git a/R/rms.R b/R/rms.R
index 58b7737..a03dd67 100644
--- a/R/rms.R
+++ b/R/rms.R
@@ -7,6 +7,9 @@
 #'
 #' Slices the audio into \code{window}-sample blocks and prints each block's
 #' overall RMS level (dB) as ametadata to \code{mfile}.
+#' @param window Window size in samples.
+#' @param mfile Path the ametadata print stream is written to.
+#' @return A single ffmpeg filter-chain string.
 #' @keywords internal
 .rms_filter <- function(window, mfile) {
     sprintf(paste0("asetnsamples=%d,astats=metadata=1:reset=1,",
@@ -19,6 +22,9 @@
 #' The print stream alternates a \code{pts_time:<t>} line and an
 #' \code{RMS_level=<db>} line per window. Digital silence prints \code{-inf},
 #' mapped to -120.
+#' @param lines Character vector of ffmpeg ametadata print output.
+#' @return A list with numeric \code{time} (window centre, seconds) and
+#'   \code{rms} (RMS level in dB).
 #' @keywords internal
 .parse_rms <- function(lines) {
     tt <- rr <- numeric(length(lines))
diff --git a/cran-comments.md b/cran-comments.md
new file mode 100644
index 0000000..7729429
--- /dev/null
+++ b/cran-comments.md
@@ -0,0 +1,28 @@
+## Submission
+
+This is a new submission.
+
+## Test environments
+
+- Local: Ubuntu 24.04, R release
+- Windows 10: R 4.6.0 and R-devel (pre-release), via `R CMD check --as-cran`
+
+## R CMD check results
+
+0 errors | 0 warnings | 1 note
+
+The one NOTE is the expected "New submission" / Maintainer note.
+
+## Notes for the reviewer
+
+- **System requirement.** The package wraps the `ffmpeg` command-line tool
+  (declared in `SystemRequirements:`, version >= 5.0). Every exported function
+  shells out to `ffmpeg`/`ffprobe`; nothing is reimplemented in R.
+
+- **`\dontrun` in examples.** All examples are wrapped in `\dontrun{}` rather
+  than `\donttest{}`. This is deliberate: the examples require both the
+  `ffmpeg` binary (which is not guaranteed to be present on the check machines)
+  and user-supplied media files (video/audio that does not ship with the
+  package). They cannot execute in CRAN's environment, which is the documented
+  case for `\dontrun` over `\donttest`. The package's tests cover the
+  command-construction logic without invoking `ffmpeg`.
diff --git a/man/align_overlaps.Rd b/man/align_overlaps.Rd
index 8414596..c19d8fd 100644
--- a/man/align_overlaps.Rd
+++ b/man/align_overlaps.Rd
@@ -25,3 +25,8 @@ each join's overlap from the positions: the seconds by which chunk
 chained chunk's duplicated head and an undersized chunk's shortfall both
 fall out of where the words actually are.
 }
+\examples{
+\dontrun{
+align_overlaps(c("media/chunk01.mp4", "media/chunk02.mp4"), "audio.mp3")
+}
+}
diff --git a/man/dot-parse_rms.Rd b/man/dot-parse_rms.Rd
index 4648f4e..e181bdd 100644
--- a/man/dot-parse_rms.Rd
+++ b/man/dot-parse_rms.Rd
@@ -5,6 +5,13 @@
 \usage{
 .parse_rms(lines)
 }
+\arguments{
+\item{lines}{Character vector of ffmpeg ametadata print output.}
+}
+\value{
+A list with numeric \code{time} (window centre, seconds) and
+  \code{rms} (RMS level in dB).
+}
 \description{
 The print stream alternates a \code{pts_time:<t>} line and an
 \code{RMS_level=<db>} line per window. Digital silence prints \code{-inf},
diff --git a/man/dot-rms_filter.Rd b/man/dot-rms_filter.Rd
index 20722d1..f34d94c 100644
--- a/man/dot-rms_filter.Rd
+++ b/man/dot-rms_filter.Rd
@@ -5,6 +5,14 @@
 \usage{
 .rms_filter(window, mfile)
 }
+\arguments{
+\item{window}{Window size in samples.}
+
+\item{mfile}{Path the ametadata print stream is written to.}
+}
+\value{
+A single ffmpeg filter-chain string.
+}
 \description{
 Slices the audio into \code{window}-sample blocks and prints each block's
 overall RMS level (dB) as ametadata to \code{mfile}.