Skip to content

The ultimate harness for coding agents and beyond. All your agents. All your machines. One command center. Start with code, then follow your curiosity and build across disciplines: CAD, circuits, robots, games and music.

refactor: rename backend release tag suffix from _api to _backend

0xken154761aa8c2ba4
diff --git a/.github/workflows/production-be-build.yaml b/.github/workflows/production-be-build.yaml
index 05bf635ed..a59a6db95 100644
--- a/.github/workflows/production-be-build.yaml
+++ b/.github/workflows/production-be-build.yaml
@@ -1,7 +1,7 @@
name: Docker production backend build
# Builds backend/Dockerfile.k8s and pushes it to gcr.io/autonomous-ecm/autonomous-code-be, on a
-# pushed `v*_api` tag. The `_api` suffix is what keeps this trigger and release.yml's `v*_cli` one
+# pushed `v*_backend` tag. The `_backend` suffix is what keeps this trigger and release.yml's `v*_cli` one
# from firing on each other now that the CLI and the backend share a repo.
#
# The build itself lives in autonomous-ai/github-templates. This repo is the only caller of the `-v2`
@@ -20,7 +20,7 @@ name: Docker production backend build
on:
push:
tags:
- - 'v*_api'
+ - 'v*_backend'
jobs:
build-autonomous-code-be-production:
diff --git a/.github/workflows/release-desktop.yml b/.github/workflows/release-desktop.yml
index 5c66b997b..6327f15b6 100644
--- a/.github/workflows/release-desktop.yml
+++ b/.github/workflows/release-desktop.yml
@@ -4,7 +4,7 @@ name: Release desktop
# (or `make release-desktop` from the repo root, which does exactly that).
#
# The `_desktop` suffix exists only so this workflow's tag trigger never fires on the CLI's
-# `v*_cli` or the backend's `v*_api` release tags, now that all three live in this one repo. It is
+# `v*_cli` or the backend's `v*_backend` release tags, now that all three live in this one repo. It is
# stripped below before anything treats the value as a version — the manifest, --build-name, the
# release title and the updater's semver comparison all only ever see a bare X.Y.Z.
#
diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml
index abe97464b..60ed6faab 100644
--- a/.github/workflows/release.yml
+++ b/.github/workflows/release.yml
@@ -3,7 +3,7 @@ name: Release
# Cut a release by pushing a tag: git tag -a v0.1.72_cli -F notes.md && git push origin v0.1.72_cli
#
# The `_cli` suffix exists only so this workflow's tag trigger never fires on the backend's
-# `v*_api` release tags (see .github/workflows/production-be-build.yaml) now that both live in one
+# `v*_backend` release tags (see .github/workflows/production-be-build.yaml) now that both live in one
# repo. It is stripped below before anything treats it as a version — the manifest, the release
# title, and the daemon's self-update comparison all only ever see a bare X.Y.Z.
#
diff --git a/Makefile b/Makefile
index 1d6edad1f..cecfa06a5 100644
--- a/Makefile
+++ b/Makefile
@@ -17,7 +17,7 @@ cli-test:
release-cli:
bash cli/scripts/release-cli.sh $(ARGS)
-## release-backend: tag this commit vX.Y.Z_api and push the tag — CI builds the backend's Docker
+## release-backend: tag this commit vX.Y.Z_backend and push the tag — CI builds the backend's Docker
## image and rolls it out. ARGS=minor|major|X.Y.Z to bump differently, ARGS=--dry-run to preview.
## See backend/scripts/release-be.sh.
release-backend:
diff --git a/backend/README.md b/backend/README.md
index 600cf3862..95bdb5d6c 100644
--- a/backend/README.md
+++ b/backend/README.md
@@ -63,7 +63,7 @@ When direct ICE fails, the terminal opens on WebSocket relay automatically.
## Deploy
`../.github/workflows/production-be-build.yaml` builds and pushes the production Docker image on
-tag push. Backend releases are tagged **`vX.Y.Z_api`** — the `_api` suffix is what routes the push
+tag push. Backend releases are tagged **`vX.Y.Z_backend`** — the `_backend` suffix is what routes the push
to this workflow instead of the CLI's `../.github/workflows/release.yml` (which only reacts to
`vX.Y.Z_cli`), now that both live in this one repo. Cut one with:
diff --git a/backend/scripts/release-be.sh b/backend/scripts/release-be.sh
index acbb4468b..cc82bb142 100755
--- a/backend/scripts/release-be.sh
+++ b/backend/scripts/release-be.sh
@@ -1,19 +1,21 @@
#!/usr/bin/env bash
-# Cut a backend release: bump the version, tag HEAD `vX.Y.Z_api`, push the tag.
+# Cut a backend release: bump the version, tag HEAD `vX.Y.Z_backend`, push the tag.
#
-# ./scripts/release-be.sh # patch bump (v1.1.2_api -> v1.1.3_api)
-# ./scripts/release-be.sh minor # v1.1.2_api -> v1.2.0_api
-# ./scripts/release-be.sh major # v1.1.2_api -> v2.0.0_api
-# ./scripts/release-be.sh 1.4.1 # exact version -> v1.4.1_api
+# ./scripts/release-be.sh # patch bump (v1.1.2_backend -> v1.1.3_backend)
+# ./scripts/release-be.sh minor # v1.1.2_backend -> v1.2.0_backend
+# ./scripts/release-be.sh major # v1.1.2_backend -> v2.0.0_backend
+# ./scripts/release-be.sh 1.4.1 # exact version -> v1.4.1_backend
# ./scripts/release-be.sh --dry-run # print what it WOULD do, touch nothing
#
# Pushing the tag is the whole point: ../.github/workflows/production-be-build.yaml triggers on
-# `vX.Y.Z_api` tags and builds Dockerfile.k8s -> autonomous-code-be.
+# `vX.Y.Z_backend` tags and builds Dockerfile.k8s -> autonomous-code-be.
# The image ArgoCD deploys is named after this tag, so the tag must point at the commit you want live.
#
-# The `_api` suffix exists only so this tag never also fires the CLI's release workflow
+# The `_backend` suffix exists only so this tag never also fires the CLI's release workflow
# (../.github/workflows/release.yml, which reacts to `vX.Y.Z_cli`) now that both live in one repo.
-# It is stripped below before anything treats it as a version.
+# It is stripped below before anything treats it as a version. Releases before 2026-09-14 used the
+# `_api` suffix (up to v1.2.15_api); the backend's version line was deliberately restarted at
+# v0.0.1_backend, so those legacy tags are ignored here and nothing builds from them any more.
# Bash 3.2-compatible (macOS default).
set -euo pipefail
@@ -58,11 +60,12 @@ if ! git merge-base --is-ancestor "$HEAD_SHA" "origin/$BRANCH" 2>/dev/null; then
fi
# --- next version: highest valid SemVer release tag, then bump.
-# Tags are "vX.Y.Z_api" — strip "v" and "_api" so LATEST is a bare X.Y.Z.
-# `sort -V` handles numeric ordering while grep drops malformed tags.
-LATEST="$(git tag -l 'v*_api' \
- | { grep -E '^v(0|[1-9][0-9]*)\.(0|[1-9][0-9]*)\.(0|[1-9][0-9]*)_api$' || true; } \
- | sed -E 's/^v//; s/_api$//' \
+# Tags are "vX.Y.Z_backend" — strip "v" and "_backend" so LATEST is a bare X.Y.Z. No tag yet
+# means 0.0.0, so the first release is v0.0.1_backend. `sort -V` handles numeric ordering while
+# grep drops malformed tags.
+LATEST="$(git tag -l 'v*_backend' \
+ | { grep -E '^v(0|[1-9][0-9]*)\.(0|[1-9][0-9]*)\.(0|[1-9][0-9]*)_backend$' || true; } \
+ | sed -E 's/^v//; s/_backend$//' \
| sort -V | tail -1)"
LATEST="${LATEST:-0.0.0}"
@@ -78,7 +81,7 @@ case "$BUMP" in
exact) NEXT="$EXACT" ;;
esac
-TAG="v${NEXT}_api"
+TAG="v${NEXT}_backend"
if git rev-parse -q --verify "refs/tags/$TAG" >/dev/null; then
echo "ERROR tag $TAG already exists. Re-tagging a released version breaks the image<->commit mapping." >&2
@@ -87,7 +90,7 @@ fi
echo " service : $SERVICE"
echo " branch : $BRANCH @ $(git rev-parse --short HEAD) $(git log -1 --format=%s | cut -c1-60)"
-echo " latest : v${LATEST}_api"
+echo " latest : v${LATEST}_backend"
echo " new tag : $TAG [$BUMP]"
if [[ "$DRY_RUN" == "1" ]]; then
diff --git a/cli/RELEASE.md b/cli/RELEASE.md
index 4ec0c5f6a..921e816bf 100644
--- a/cli/RELEASE.md
+++ b/cli/RELEASE.md
@@ -5,7 +5,7 @@ tag: `.github/workflows/release.yml` bundles the CLI, publishes it to that bucke
Release. **The tag IS the version — CI never bumps.**
Tags are `vX.Y.Z_cli`, not plain `vX.Y.Z`: the backend now lives in this same repo under `backend/`
-and releases as `vX.Y.Z_api` (see `../backend/README.md`), so the suffix is what tells each tag-push
+and releases as `vX.Y.Z_backend` (see `../backend/README.md`), so the suffix is what tells each tag-push
trigger which workflow to run. It's stripped before anything treats it as a version — the published
version everywhere below (manifest, release title, `ADAPTER_VERSION`) is always a bare `X.Y.Z`.
diff --git a/cli/scripts/release-cli.sh b/cli/scripts/release-cli.sh
index 1fd300a36..12e192f07 100755
--- a/cli/scripts/release-cli.sh
+++ b/cli/scripts/release-cli.sh
@@ -3,7 +3,7 @@
# .github/workflows/release.yml bundles the CLI, publishes it to GCS, and creates the GitHub Release.
# The tag IS the version; CI never bumps on its own. Tags are "vX.Y.Z_cli" — the "_cli" suffix is
# stripped by release.yml before anything treats it as a version; it exists only so this tag never
-# also triggers the backend's "vX.Y.Z_api" release (.github/workflows/production-be-build.yaml).
+# also triggers the backend's "vX.Y.Z_backend" release (.github/workflows/production-be-build.yaml).
#
# Usage:
# bash cli/scripts/release-cli.sh # bump the patch and release
@@ -107,7 +107,7 @@ fi
# --- current version, source 1: the highest release tag in git ---
# The brace group keeps `set -e` from killing the script when grep legitimately matches nothing.
# Tags are "vX.Y.Z_cli" — the suffix only exists so release.yml's tag trigger never collides with
-# the backend's "vX.Y.Z_api" tags now that both live in one repo. Strip "v" and "_cli" so TAG_VER
+# the backend's "vX.Y.Z_backend" tags now that both live in one repo. Strip "v" and "_cli" so TAG_VER
# is always a bare X.Y.Z for the version-comparison logic below.
TAG_VER="$(git tag -l 'v*_cli' \
| { grep -E '^v(0|[1-9][0-9]*)\.(0|[1-9][0-9]*)\.(0|[1-9][0-9]*)_cli$' || true; } \
diff --git a/desktop/RELEASE.md b/desktop/RELEASE.md
index 3a095a1ed..02e3bd2e7 100644
--- a/desktop/RELEASE.md
+++ b/desktop/RELEASE.md
@@ -6,7 +6,7 @@ Apple Silicon on Impeller — see "Two macOS builds" below) and both Linux archi
to that bucket, and cuts the GitHub Release. **The tag IS the version — CI never bumps.**
Tags are `vX.Y.Z_desktop`, not plain `vX.Y.Z`: the CLI (`cli/`, tags `vX.Y.Z_cli`) and the backend
-(`backend/`, tags `vX.Y.Z_api`) release from this same repo, so the suffix is what tells each
+(`backend/`, tags `vX.Y.Z_backend`) release from this same repo, so the suffix is what tells each
tag-push trigger which workflow to run. It is stripped before anything treats it as a version — the
manifest, `--build-name`, the release title and the updater's comparison all see a bare `X.Y.Z`.
diff --git a/desktop/scripts/release-desktop.sh b/desktop/scripts/release-desktop.sh
index 4e1e4f160..77e0b3012 100755
--- a/desktop/scripts/release-desktop.sh
+++ b/desktop/scripts/release-desktop.sh
@@ -3,7 +3,7 @@
# ../.github/workflows/release-desktop.yml builds macOS + Linux, publishes to GCS, and creates the
# GitHub Release. The tag IS the version; CI never bumps on its own. Tags are "vX.Y.Z_desktop" — the
# "_desktop" suffix is stripped by release-desktop.yml before anything treats it as a version; it
-# exists only so this tag never also triggers the CLI's "vX.Y.Z_cli" or the backend's "vX.Y.Z_api"
+# exists only so this tag never also triggers the CLI's "vX.Y.Z_cli" or the backend's "vX.Y.Z_backend"
# release, now that all three live in one repo.
#
# Usage (from the repo root, `make release-desktop ARGS=...` runs the same thing):
diff --git a/docs/cicd.md b/docs/cicd.md
index 13b3c6167..1c1e5615d 100644
--- a/docs/cicd.md
+++ b/docs/cicd.md
@@ -7,7 +7,7 @@ app (`desktop/`) — and every workflow lives under [`.github/workflows/`](../.g
|---|---|---|
| `ci.yml` | every pull request against `main`; on demand | `npm ci` → `npx tsc --noEmit` → `npx vitest run`, i.e. `make cli-test` |
| `release.yml` | a pushed `v*_cli` tag (`make release-cli`); on demand | bundles and publishes the CLI, then verifies the published bytes |
-| `production-be-build.yaml` | a pushed `v*_api` tag (`make release-backend`) | builds `backend/Dockerfile.k8s` and pushes it to GCR |
+| `production-be-build.yaml` | a pushed `v*_backend` tag (`make release-backend`) | builds `backend/Dockerfile.k8s` and pushes it to GCR |
| `release-desktop.yml` | a pushed `v*_desktop` tag (`make release-desktop`); on demand | builds, signs and publishes both macOS builds and both Linux architectures of the desktop app — see `desktop/RELEASE.md` |
| `desktop-internal-build.yml` | a push to `internal/**`; on demand | a signed, notarized desktop build for testers behind an unlisted link, no release |
@@ -22,7 +22,7 @@ specific to this repo.
## Why the tag suffixes
Three release triggers live in one repo, so each one has to be unmistakable: `v0.1.72_cli` publishes
-the CLI, `v1.4.0_api` builds the backend image, and `v1.1.14_desktop` releases the desktop app. Each
+the CLI, `v1.4.0_backend` builds the backend image, and `v1.1.14_desktop` releases the desktop app. Each
workflow (and each `release-*.sh`) strips both the `v` and its suffix before anything treats the
string as a version — the suffix is a routing marker for the trigger and never reaches the manifest,
the release title or an updater's version comparison.
@@ -191,7 +191,7 @@ Two identities rather than one is not ceremony here: it means a compromised imag
the release bucket that every daemon on the fleet polls, and vice versa.
**Merge order matters.** `docker-build-and-push-v2.yaml` has to exist on `github-templates`' `main`
-before this repo's caller points at it. The reverse order breaks every `v*_api` tag build until it is
+before this repo's caller points at it. The reverse order breaks every `v*_backend` tag build until it is
fixed — though only this repo's, which is the point of the clone.
**`permissions:` belongs to the caller.** A called workflow can never hold more permission than the