refactor: retire the fabledcurator-ml image name entirely (4311)

Operator, 2026-09-23: *"we don't need to keep this as I'll be collapsing my
stack to the same consolidated version that we're building here."*

That was the only thing holding the name up. `fabledcurator-ml` has been the
same bytes as `fabledcurator` since milestone 422 step 6, and 8152684 had
already stopped rebuilding it — this removes the name.

Gone: the whole `build-ml` job (137 more lines), its entry in promote's
`needs` and in promote's tag loop, `ML_PATHS` and the `ml)` case in
artifacts.sh, the `ml` artifact in test_artifact_paths, the image in
release_notes.py and its test, and the README row. `docker-compose.yml`'s
ml-worker service now runs `fabledcurator:latest` with `command:
["ml-worker"]` — the service still exists for anyone who wants lane
separation, it just no longer needs a second image name.

Prose that said "all three images" is now "both", except where it means
three VALUES (revision/version/epoch) or records what happened on 2026-08-30,
when there genuinely were three.

Published tags are left alone. Nothing new goes to that name; what is already
in the registry stays pullable, so a stack that has not moved yet keeps
running rather than losing its image mid-flight.

## A latent trap this made reachable, found by running the script

`artifacts.sh revision ml` did not fail. It printed the usage line to stderr
and answered with **the newest commit in the whole repository** — a
real-looking 12-char sha on stdout, exit 0.

`newest()` inlined the path set as `git log ... -- $(cmd_paths "$1")`, and
`usage` exits from the command SUBSHELL, so the substitution came back empty
and `git log HEAD --` had no pathspec left to filter by. The reuse check
would have compared that answer against a published label, missed, and
rebuilt on every push forever with nothing going red.

It could not be fixed inside `newest`, which was my first attempt: every
caller wraps it in a substitution too, so its `exit` also died in a subshell —
stdout went empty but the status stayed 0. The guard has to run in the main
shell, so it validates the artifact name at dispatch. `newest` keeps a
defensive `|| exit 2` for a future caller that reaches it another way, and
the comment says which of the two is the real check.

Latent while every name callers passed was valid. Removing `ml` from the set
is what made a name that used to work start taking the silent path, so the
test that pins it lands here.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LVjrnpQjRgHdvq95rASoiR

.forgejo/workflows/build.yml

@@ -10,9 +10,12 @@ on:
     # pressure to merge in order to try something does not come from
     # carelessness; it comes from `:dev` being unable to carry the build.
     #
-    # All three images build on dev, deliberately: a `:dev` web image paired
+    # BOTH images build on dev, deliberately: a `:dev` web image paired
-    # with a stale `:dev` ml or agent is a worse trap than no dev channel at
+    # with a stale `:dev` agent is a worse trap than no dev channel at all,
-    # all, since the mismatch only shows up as a runtime failure.
+    # since the mismatch only shows up as a runtime failure. (It was three
+    # until #4311 — `fabledcurator-ml` was the same bytes as `fabledcurator`
+    # under a second name, and the stack that needed the second name is being
+    # collapsed onto the consolidated image.)
     branches: [main, dev]
     #
     # NO tag trigger (milestone 318 step 2). A `v*` tag names a commit `main`
@@ -35,8 +38,8 @@ on:
   # workflow is not shipped bytes, so it is in no artifact's path set. Putting
   # it in one would re-version every artifact for a comment change.
   #
-  # ONE input, not one per artifact. Forcing all three is cheap once the
+  # ONE input, not one per artifact. Forcing both is cheap once the registry
-  # registry cache is warm (#3114), and three booleans is an interface nobody
+  # cache is warm (#3114), and a boolean per artifact is an interface nobody
   # remembers the meaning of.
   workflow_dispatch:
     inputs:
@@ -787,8 +790,8 @@ jobs:
           #
           # build-web additionally exposes this as `outputs.candidate`, which is
           # what gates the `promote` job — a job's `if:` cannot read `env`, and
-          # one flag is enough because all three derive it from the same
+          # one flag is enough because both jobs derive it from the same
-          # IS_REFRESH. ml and agent do not re-emit it; a second copy nothing
+          # IS_REFRESH. build-agent does not re-emit it; a second copy nothing
           # reads is the kind of thing that later reads as load-bearing.
           if [ "${IS_REFRESH:-}" = "true" ]; then
             echo "build_ref=$IMAGE:refresh-candidate" >> "$GITHUB_OUTPUT"
@@ -1523,18 +1526,19 @@ jobs:
   # a FAILED gate blocks would have published unverified images while reporting
   # success. Not running is not the same as passing.
   #
-  # All three images promote TOGETHER, or none do. They are one stack: build.yml
+  # BOTH images promote TOGETHER, or neither does. They are one stack:
-  # already refuses to publish a :dev web image beside a stale :dev ml, because
+  # build.yml already refuses to publish a :dev web image beside a stale :dev
-  # the mismatch only shows up as a runtime failure. A refresh that published ml
+  # agent, because the mismatch only shows up as a runtime failure. A refresh
-  # and withheld web would be that same trap, arrived at through the gate.
+  # that published the agent and withheld web would be that same trap, arrived
+  # at through the gate.
   #
   # The gate covers the web image only (milestone 362 step 3 scoped it there),
-  # so ml and agent are being held to web's verdict rather than their own. That
+  # so the agent is being held to web's verdict rather than its own. That is
-  # is deliberate and it is the conservative direction — they ship together, so
+  # deliberate and it is the conservative direction — they ship together, so
-  # the weakest evidence should govern all three — but it is not the same as
+  # the weakest evidence should govern both — but it is not the same as having
-  # having smoked them, and it should not be read as if it were.
+  # smoked it, and it should not be read as if it were.
   promote:
-    needs: [build-web, build-ml, build-agent, smoke-web]
+    needs: [build-web, build-agent, smoke-web]
     # Only a refresh publishes through a candidate; a push writes its channel
     # tag directly from the build. Reads the same reuse-step decision the build
     # took, via a job output — a job's `if:` cannot see the `env` context.
@@ -1555,7 +1559,7 @@ jobs:
           TAG=latest
           FAILED=""
 
-          for NAME in fabledcurator fabledcurator-ml fabledcurator-agent; do
+          for NAME in fabledcurator fabledcurator-agent; do
             REPO="bvandeusen/$NAME"
             echo "promote: $REPO"
 
@@ -1616,144 +1620,7 @@ jobs:
             echo "promote: idempotent." >&2
             exit 1
           fi
-          echo "promote: all three channel tags moved"
+          echo "promote: both channel tags moved"
-  # NOT A BUILD. `fabledcurator-ml` is the SAME IMAGE as `fabledcurator` and
-  # has been since milestone 422 step 6 merged the ML layers into the one
-  # Dockerfile — this job built `file: Dockerfile, context: .`, byte for byte
-  # what build-web builds, and published it under a second name.
-  #
-  # So it was doing the whole build twice. Measured on run 7282, a cold cache:
-  # build-web 1m54s, build-ml 1m55s, for identical output — plus a second
-  # push of a few hundred MB. Operator, 2026-09-22: *"please fix the CI so it
-  # doesn't do this superfluous work."*
-  #
-  # It now publishes NOTHING OF ITS OWN. It re-tags the manifest build-web
-  # already pushed, so the second name goes on working for the operator's
-  # Swarm stack — which still references it — while CI stops building twice.
-  #
-  # The name is retired entirely in #4311, once that stack points its
-  # ml-worker service at `fabledcurator:latest` with `command: ["ml-worker"]`.
-  # Nothing here is load-bearing after that: the whole job goes, rather than
-  # this comment growing another paragraph.
-  build-ml:
-    # Was parallel with build-web. It cannot be any more — there is nothing to
-    # copy until build-web has decided what this run publishes. That is the
-    # cost of not building twice, and it is seconds: this job transfers no
-    # layers the runner does not already have.
-    needs: [build-web]
-    runs-on: python-ci
-    container:
-      image: git.fabledsword.com/bvandeusen/ci-python:3.14
-    steps:
-      - uses: actions/checkout@v4
-        with:
-          ref: ${{ env.BUILD_REF }}
-          # Shallow is enough now. This job derives no version from history:
-          # it publishes no content, so it has no artifact identity of its own
-          # and `artifacts.sh revision ml` decides nothing here any more.
-
-      # See sign-extension's copy for why this guard exists.
-      - name: Guard — a scheduled run must have checked out main
-        if: env.IS_REFRESH == 'true'
-        run: |
-          set -eu
-          BRANCH=$(git rev-parse --abbrev-ref HEAD)
-          echo "schedule: HEAD is $BRANCH ($(git rev-parse --short HEAD))"
-          if [ "$BRANCH" != "main" ]; then
-            echo "schedule: expected main, got '$BRANCH'." >&2
-            echo "schedule: BUILD_REF was not honoured by the runner." >&2
-            echo "schedule: refusing to publish a channel tag from it." >&2
-            exit 1
-          fi
-
-      # Shell step rather than docker/login-action — see build-web's note on
-      # the shared action-cache race (#3118).
-      - name: Login to Forgejo registry
-        env:
-          TOKEN: ${{ secrets.RELEASE_TOKEN }}
-          ACTOR: ${{ github.actor }}
-        run: echo "$TOKEN" | docker login git.fabledsword.com -u "$ACTOR" --password-stdin
-
-      - name: Point fabledcurator-ml at the image fabledcurator published
-        env:
-          SRC: git.fabledsword.com/bvandeusen/fabledcurator
-          DST: git.fabledsword.com/bvandeusen/fabledcurator-ml
-          BUILT_DIGEST: ${{ needs.build-web.outputs.digest }}
-          PUBLISHED_DIGEST: ${{ needs.build-web.outputs.published_digest }}
-        run: |
-          set -eu
-          # WHERE THE BYTES COME FROM. build-web either pushed a manifest this
-          # run, or it hit reuse — in which case the digest its channel tag
-          # already names is the right source, because that is what "hit"
-          # MEANS: the reuse step read this commit's fc.revision off that tag.
-          DIGEST="${BUILT_DIGEST:-}"
-          [ -n "$DIGEST" ] || DIGEST="${PUBLISHED_DIGEST:-}"
-          if [ -z "$DIGEST" ]; then
-            echo "alias: FAILED — build-web neither built an image nor" >&2
-            echo "alias: resolved a published one, so there is nothing for" >&2
-            echo "alias: fabledcurator-ml to be pointed at." >&2
-            exit 1
-          fi
-
-          # WHICH TAGS, mirroring exactly what build-web wrote for itself.
-          #
-          # A refresh writes the CANDIDATE tag and nothing else, so `promote`
-          # moves it to :latest after smoke-web passes. That keeps this name
-          # under the same gate it has today and needs no change to promote,
-          # which already loops over all three image names.
-          SHORT_SHA=$(printf '%s' "$GITHUB_SHA" | cut -c1-7)
-          if [ "${IS_REFRESH:-}" = "true" ]; then
-            TAGS="refresh-candidate"
-          elif [ "${GITHUB_REF##*/}" = "main" ]; then
-            TAGS="latest c-${SHORT_SHA}"
-          else
-            TAGS="dev"
-          fi
-
-          # Pull, tag, push — deliberately NOT `imagetools create`.
-          #
-          # That wraps its source in an INDEX, and `.Image.Config.Labels` does
-          # not resolve through one (#3183, run 4751). Worse here than it was
-          # there: promote asks the registry for IMAGE manifest media types
-          # ONLY, so an index sitting at :refresh-candidate would fail its
-          # `test -n "$CT"` and break the weekly refresh for this name.
-          #
-          # MEASURED on run 7300, the first execution: the push moved ZERO
-          # layer bytes. Every blob came back either "Layer already exists" or
-          # "Mounted from bvandeusen/fabledcurator" — the registry does
-          # cross-repository blob mounts between two repos of one owner, so
-          # this copies a manifest and nothing else. 22s, against the ~2min
-          # the duplicate build cost (run 7282: build-ml 1m55s).
-          #
-          # The pushed manifest gets its OWN digest (b979b145… where the
-          # source was aca4c75…) because docker re-serialises it. Same layers,
-          # same config, same fc.revision; only the manifest bytes differ.
-          # Nothing compares the two, and promote reads whatever is at the tag
-          # — but it is worth saying, because the digests looking different is
-          # the first thing someone will notice and misread as a second build.
-          docker pull "$SRC@$DIGEST"
-          for t in $TAGS; do
-            echo "alias: $DST:$t -> $SRC@$DIGEST"
-            docker tag "$SRC@$DIGEST" "$DST:$t"
-            docker push "$DST:$t"
-          done
-
-          # Read it back. A push that reported success but left the tag
-          # elsewhere is exactly the silent-and-plausible failure this
-          # pipeline keeps producing, and the check costs one request.
-          for t in $TAGS; do
-            NOW=$(docker buildx imagetools inspect "$DST:$t" \
-              --format '{{ index .Image.Config.Labels "fc.revision" }}' 2>/dev/null || echo "")
-            echo "alias: $DST:$t now carries fc.revision=${NOW:-<none>}"
-            if [ -z "$NOW" ]; then
-              echo "alias: FAILED — $DST:$t has no readable fc.revision." >&2
-              echo "alias: The label resolves through a plain image manifest" >&2
-              echo "alias: and not through an index, so this means the tag is" >&2
-              echo "alias: index-wrapped — which breaks promote's refresh path." >&2
-              exit 1
-            fi
-          done
-
   build-agent:
     runs-on: python-ci
     container:
@@ -1961,8 +1828,8 @@ jobs:
           #
           # build-web additionally exposes this as `outputs.candidate`, which is
           # what gates the `promote` job — a job's `if:` cannot read `env`, and
-          # one flag is enough because all three derive it from the same
+          # one flag is enough because both jobs derive it from the same
-          # IS_REFRESH. ml and agent do not re-emit it; a second copy nothing
+          # IS_REFRESH. build-agent does not re-emit it; a second copy nothing
           # reads is the kind of thing that later reads as load-bearing.
           if [ "${IS_REFRESH:-}" = "true" ]; then
             echo "build_ref=$IMAGE:refresh-candidate" >> "$GITHUB_OUTPUT"

README.md

@@ -260,12 +260,11 @@ commits since the previous tag; it builds no image.
 
 ## What's in here
 
-Five deployable pieces, built by `.forgejo/workflows/build.yml`:
+Four deployable pieces, built by `.forgejo/workflows/build.yml`:
 
 | Piece | Built from | Image | Role |
 | --- | --- | --- | --- |
-| **Web / workers** | `Dockerfile` | `fabledcurator` | Quart API + the built Vue SPA in one image. `entrypoint.sh` picks the role: `web`, `worker`, `scheduler`. The `maintenance-long` service is a second `worker` pinned to the long-running maintenance queue. |
+| **Web / workers** | `Dockerfile` | `fabledcurator` | Quart API + the built Vue SPA in one image. `entrypoint.sh` picks the role: `web`, `worker`, `scheduler`, `ml-worker`, or `all` (every lane under supervisord, the single-container layout). The `maintenance-long` service is a second `worker` pinned to the long-running maintenance queue. |
-| **ML worker** | `Dockerfile` | `fabledcurator-ml` | The same image as the web service since milestone 422 — one image serves every lane. Published under this name too, for stacks that still reference it. |
 | **GPU agent** | `agent/Dockerfile` | `fabledcurator-agent` | Optional desktop-GPU worker (`agent/`). Leases jobs over **HTTP only** — never touches the database or Redis. See `agent/README.md`. |
 | **Firefox extension** | `extension/` | signed XPI | MV3 extension: pushes platform session cookies into FC and adds a creator as a Source in one click. AMO-signed on both `dev` and `main` (one signature per extension change, shared by the two channels), bundled into that channel's web image and served from Settings → Maintenance. See `extension/README.md`. |
 | **Data** | — | `pgvector/pgvector:pg16`, `redis:7-alpine` | Postgres with pgvector for embeddings; Redis as the Celery broker. |

docker-compose.yml

@@ -233,7 +233,7 @@ services:
       redis:    { condition: service_healthy }
 
   ml-worker:
-    image: git.fabledsword.com/bvandeusen/fabledcurator-ml:latest
+    image: git.fabledsword.com/bvandeusen/fabledcurator:latest
     command: ["ml-worker"]
     # A single GPU inference pass can run tens of seconds — let it finish.
     stop_grace_period: 120s

scripts/artifacts.sh

@@ -8,7 +8,7 @@
 # never reaches the image — see DERIVER below, and #3156 for the same finding
 # about packaging.sh.
 #
-# Four artifacts, four independent versions. An artifact whose shipped files
+# Three artifacts, three independent versions. An artifact whose shipped files
 # did not change keeps its version and does not rebuild — that is the whole
 # point, and it is why each path set must match its Dockerfile rather than
 # being a plausible guess. Getting a set wrong is quiet in BOTH directions:
@@ -51,21 +51,6 @@ ROOT=$(git rev-parse --show-toplevel)
 # rather than restated — one definition, per #2397.
 WEB_PATHS='Dockerfile requirements.txt requirements-ml.txt backend alembic alembic.ini entrypoint.sh frontend :(exclude)frontend/test :(exclude)frontend/test/**'
 
-# ml — THE SAME IMAGE as web since milestone 422 step 6, built from the same
-# Dockerfile with the same context and published under a second name only
-# because the operator's Swarm stack still references it.
-#
-# So it is the same path set, by assignment rather than by a copy that would
-# drift. This was two lists describing one image for about an hour, and the
-# artifact-paths guard caught it immediately: the merged Dockerfile carries
-# the frontend-builder stage, so the ml image copies `frontend/package.json`,
-# and the old ml list did not cover it — its version would not have moved when
-# the frontend changed, and a pinned build would serve stale bytes (#3202).
-#
-# `requirements-ml.txt` is in WEB_PATHS for the mirror-image reason: the web
-# image now installs the ML requirements, so changing them changes it.
-ML_PATHS="$WEB_PATHS"
-
 # agent (agent/Dockerfile, context `agent`) — copies requirements.txt and
 # fc_agent only. agent/README.md, agent/docker-compose.yml and agent/ruff.toml
 # live in the directory but never reach the image, so they must not re-version
@@ -78,7 +63,7 @@ AGENT_PATHS='agent/Dockerfile agent/requirements.txt agent/fc_agent'
 #
 # Why web and nothing else. Every artifact stamps `fc.revision`, but only web
 # also stamps a version (build.yml line ~488 feeds `version web` to the
-# FC_VERSION build arg; ml and agent ask for `revision` alone, and the
+# FC_VERSION build arg; the agent asks for `revision` alone, and the
 # extension takes its version from packaging.sh). For a revision-only artifact
 # this file needs no entry: any change to how the revision is COMPUTED changes
 # the derived value, which then disagrees with the label on the published image
@@ -101,7 +86,7 @@ DERIVER='scripts/artifacts.sh'
 
 
 usage() {
-  echo "usage: artifacts.sh {paths|revision|version|epoch} {web|ml|agent|extension}" >&2
+  echo "usage: artifacts.sh {paths|revision|version|epoch} {web|agent|extension}" >&2
   exit 2
 }
 
@@ -115,9 +100,6 @@ ext_paths() {
 cmd_paths() {
   case "$1" in
     web)       echo "$WEB_PATHS $DERIVER $(ext_paths)" ;;
-    # Identical to web, deliberately: it IS web's image. That includes the
-    # deriver and the bundled extension — the XPI is in these bytes too.
-    ml)        echo "$ML_PATHS $DERIVER $(ext_paths)" ;;
     agent)     echo "$AGENT_PATHS" ;;
     extension) ext_paths ;;
     *)         usage ;;
@@ -128,8 +110,21 @@ cmd_paths() {
 # Unquoted on purpose: the pathspec must word-split into separate args.
 # Globbing is already off script-wide.
 newest() {
-  # shellcheck disable=SC2046
+  # Resolve the path set into a variable rather than inlining it as
-  set -- "$(cd "$ROOT" && git log --format='%ct %H' HEAD -- $(cmd_paths "$1") \
+  # `git log ... -- $(cmd_paths "$1")`. An empty substitution there leaves
+  # `git log HEAD --` with NO pathspec, which answers with the newest commit
+  # in the whole repository instead of failing — see the dispatch guard at the
+  # foot of this file for what that cost and why the real check lives there.
+  #
+  # This `|| exit 2` cannot be that check (it exits a subshell, since every
+  # caller wraps `newest` in one) — it is belt and braces for a future caller
+  # that reaches `newest` without passing the guard.
+  paths=$(cmd_paths "$1") || exit 2
+
+  # Unquoted on purpose: the pathspec must word-split into separate args.
+  # Globbing is off script-wide (`set -f`).
+  # shellcheck disable=SC2086
+  set -- "$(cd "$ROOT" && git log --format='%ct %H' HEAD -- $paths \
             | sort -n | tail -1)"
   if [ -z "$1" ]; then
     echo "artifacts.sh: no commit touches this artifact's shipped files" >&2
@@ -226,6 +221,28 @@ cmd_version() {
 }
 
 [ $# -ge 2 ] || usage
+
+# Validate the ARTIFACT here, in the main shell, before anything dispatches.
+#
+# It cannot be done deeper down. `cmd_revision`, `cmd_version` and `cmd_epoch`
+# all call `newest` inside a command substitution, and `newest` resolves the
+# path set inside another one — so an `exit` from either kills only that
+# subshell. The script carried on, printed nothing, and exited 0.
+#
+# Before this, `revision ml` was worse than nothing: the empty substitution
+# left `git log HEAD --` with no pathspec, so it answered with the newest
+# commit in the WHOLE REPOSITORY — a real-looking 12-char sha on stdout, exit
+# 0, and the usage line on stderr where no caller reads it. The reuse check
+# would have compared that against a published label, missed, and rebuilt
+# every push forever without a red run anywhere.
+#
+# Latent while every name callers passed was valid. #4311 removed `ml` from
+# the set, which made a name that used to work start taking that path.
+case "$2" in
+  web|agent|extension) ;;
+  *) usage ;;
+esac
+
 case "$1" in
   paths)    cmd_paths "$2" ;;
   revision) cmd_revision "$2" ;;

scripts/release_notes.py

@@ -83,7 +83,6 @@ API = "https://git.fabledsword.com/api/v1/repos/bvandeusen/FabledCurator"
 
 IMAGES = (
     "git.fabledsword.com/bvandeusen/fabledcurator",
-    "git.fabledsword.com/bvandeusen/fabledcurator-ml",
     "git.fabledsword.com/bvandeusen/fabledcurator-agent",
 )
 

tests/test_artifact_paths.py

@@ -31,7 +31,6 @@ ROOT = Path(__file__).resolve().parent.parent
 # artifact -> (dockerfile, build context relative to the repo root)
 ARTIFACTS = {
     "web": ("Dockerfile", ""),
-    "ml": ("Dockerfile", ""),
     "agent": ("agent/Dockerfile", "agent"),
 }
 
@@ -133,7 +132,7 @@ DERIVERS = [
     # orphaned. Guarded for web too, since web bundles what the extension makes.
     ("extension/scripts/packaging.sh", ("extension", "web")),
     # artifacts.sh decides the FC_VERSION baked into the web image (#3202).
-    # Web only, and deliberately: ml and agent ask this script for `revision`
+    # Web only, and deliberately: the agent asks this script for `revision`
     # alone, so they are covered by the self-correcting path above, and the
     # extension takes its version from packaging.sh. Milestone 318 step 5 is
     # the worked instance — b3989d0 and 5771fd5 share revision fb2c4d5b80be
@@ -197,3 +196,30 @@ def test_files_that_never_reach_an_image_do_not_version_it(artifact, path):
         f"image — it would re-version and rebuild {artifact} for a change it "
         f"does not ship."
     )
+
+
+def test_an_unknown_artifact_fails_instead_of_answering():
+    """It used to answer — with the newest commit in the whole repository.
+
+    `newest()` inlined the path set as `git log ... -- $(cmd_paths "$1")`.
+    `usage` exits from the command SUBSHELL, so an unknown name made the
+    substitution come back empty and `git log HEAD --` walked everything:
+    stdout got a real-looking 12-char sha, the exit code was 0, and the usage
+    line went to stderr where no caller reads it. A wrong answer shaped
+    exactly like a right one, which the reuse check would have compared
+    against a published label and quietly rebuilt on forever.
+
+    Latent while every name callers passed was valid. #4311 removed `ml` from
+    that set, which is what made a name that used to work start taking the
+    silent path.
+    """
+    out = subprocess.run(
+        ["sh", str(ROOT / "scripts" / "artifacts.sh"), "revision", "ml"],
+        capture_output=True, text=True, cwd=ROOT,
+    )
+    assert out.returncode != 0, (
+        f"an unknown artifact exited 0 and printed {out.stdout!r}"
+    )
+    assert not out.stdout.strip(), (
+        f"an unknown artifact printed {out.stdout!r} on stdout"
+    )

tests/test_release_notes.py

@@ -195,13 +195,13 @@ def test_a_non_tag_ref_renders_but_refuses_to_claim_it_published():
     body_of(out)
 
 
-def test_the_rollback_refs_name_all_three_images():
+def test_the_rollback_refs_name_both_images():
-    """Rule 145: `:c-<sha>` is the rollback unit, and the three images move
+    """Rule 145: `:c-<sha>` is the rollback unit, and both images move
     together. A release listing only the web image sends an operator into a
-    rollback that leaves ml and agent on the newer build — the exact mismatch
+    rollback that leaves the agent on the newer build — the exact mismatch
-    build.yml builds all three on every push to avoid."""
+    build.yml builds both on every push to avoid."""
     body = body_of(notes("HEAD"))
-    for image in ("fabledcurator", "fabledcurator-ml", "fabledcurator-agent"):
+    for image in ("fabledcurator", "fabledcurator-agent"):
         assert f"bvandeusen/{image}:c-" in body, f"{image} missing from the rollback refs"