feat: add Docker CI to build all-in-one image with agent-server + automation + frontend (#634)

* feat: add Docker CI to build all-in-one image with agent-server + automation + frontend

Adds a GitHub Actions workflow (.github/workflows/docker.yml) that builds and
publishes ghcr.io/openhands/agent-canvas — a single Docker image combining:

  1. Agent Server (ghcr.io/openhands/agent-server base image from SDK repo)
  2. Automation server (pip-installed from openhands-automation)
  3. agent-canvas frontend (static build from this repo)

The automation server is pip-installed rather than copied from its Docker image
because both services share openhands-sdk, fastapi, uvicorn, pydantic, httpx
etc. — installing into the agent-server's Python 3.13 deduplicates all shared
packages. Only automation-specific deps (asyncpg, sqlalchemy, boto3, …) are
added on top.

An entrypoint script starts all three services and a static-server proxy that
unifies them behind a single port (default 8000):
  /api/automation/* → automation backend (:18001)
  /api/*            → agent-server (:18000)
  /*                → static frontend + SPA fallback

Workflow triggers:
  - Push to main: builds and pushes with branch + SHA tags
  - v* tags (releases): also pushes semver tags (1.2.3, 1.2, 1, latest)
  - PRs: builds, pushes SHA-tagged image, updates PR description with
    pull/run instructions (same pattern as the SDK repo)
  - workflow_dispatch: supports overriding base image and automation version

Files added:
  - docker/Dockerfile (multi-stage: frontend build + agent-server base)
  - docker/entrypoint.sh (process manager for all three services)
  - .dockerignore
  - .github/workflows/docker.yml

Co-authored-by: openhands <openhands@all-hands.dev>

* fix: build multi-arch Docker images (amd64 + arm64)

Adds QEMU setup for cross-compilation and defaults the platform matrix
to linux/amd64,linux/arm64 so the image works on both Intel and Apple
Silicon machines.

Co-authored-by: openhands <openhands@all-hands.dev>

* refactor: rewrite Docker workflow to match SDK repo structure

Replace the single-job QEMU approach with the same architecture-matrix
pattern used by the SDK repo's server.yml:

  1. build-and-push-image — matrix over {amd64, arm64} with native runners
     (ubuntu-24.04 for amd64, ubuntu-24.04-arm for arm64). Each job pushes
     arch-suffixed tags (e.g. sha-abc1234-amd64) and uploads build-info
     artifacts.

  2. merge-manifests — downloads both arch build-infos, strips the -amd64
     suffix from amd64 tags to derive manifest tags, and creates multi-arch
     manifests via `docker buildx imagetools create`.

  3. consolidate-build-info — aggregates all build-info and manifest-info
     artifacts into a single JSON summary (PR-only).

  4. update-pr-description — renders the summary into the PR body between
     AGENT_CANVAS_DOCKER_START/END markers.

Native runners avoid the 3-5× slowdown of QEMU emulation for arm64
builds.

Co-authored-by: openhands <openhands@all-hands.dev>

* fix: sanitize branch names in Docker tags (/ is not allowed)

Branch names like 'feat/docker-ci' produce invalid Docker tags because
'/' is forbidden in tag names. Replace '/' with '-' so the tag becomes
'feat-docker-ci-amd64'.

Co-authored-by: openhands <openhands@all-hands.dev>

* fix: default automation to SQLite and fix wait blocking proxy startup

Two bugs:

1. The automation server defaults to PostgreSQL on localhost, which
   doesn't exist in the all-in-one container. Default AUTOMATION_DB_URL
   to sqlite+aiosqlite:// so it works out of the box. Users can override
   with a real Postgres URL for production.

2. The bare 'wait' command waited for ALL background children — including
   the long-running agent-server and automation processes — so the
   static-server/proxy on port 8000 never started. Fix by waiting only
   for the wait_for_port subshell PIDs.

Verified locally: all three services start, endpoints respond correctly,
no more scheduler ConnectionRefusedError.

Co-authored-by: openhands <openhands@all-hands.dev>

* feat: add VOLUME directives for persistence and project mounts

Declare /home/openhands/.openhands (settings, secrets, conversations,
automation SQLite DB) and /projects (user code) as Docker volumes so
data survives container restarts by default. Users should bind-mount
these for durable persistence:

  docker run -v ~/.openhands:/home/openhands/.openhands \
             -v ~/projects:/projects \
             -p 8000:8000 ghcr.io/openhands/agent-canvas

Co-authored-by: openhands <openhands@all-hands.dev>

* fix: set OH_SECRET_KEY default and pre-create persistence dirs

Three issues fixed:

1. OH_SECRET_KEY was not set → agent-server refused to return encrypted
   secrets → conversation creation failed with 503. Set the same static
   default used by dev-safe.mjs / dev-docker.mjs.

2. Persistence dirs (conversations, bash_events, automation DB) were not
   pre-created → the openhands user got PermissionError when the VOLUME
   directive created them as root. Pre-create with correct ownership
   before the USER switch in the Dockerfile.

3. Set OH_PERSISTENCE_DIR, OH_CONVERSATIONS_PATH, OH_BASH_EVENTS_DIR
   defaults in the entrypoint (matching dev-docker.mjs) so data lands
   under the well-known ~/.openhands tree.

Verified locally: all three services start clean, no warnings about
OH_SECRET_KEY, SQLite migrations apply successfully.

Co-authored-by: openhands <openhands@all-hands.dev>

* chore: merge main and remove stale dev-docker.mjs references

Main removed scripts/dev-docker.mjs (Docker is no longer a dependency of
the npm package flow). Update comments in docker.yml, entrypoint.sh, and
AGENTS.md that referenced the deleted file.

Co-authored-by: openhands <openhands@all-hands.dev>

* feat: centralize config into config/defaults.json (single source of truth)

All version pins, port defaults, persistence paths, package names, and
the dev secret key now live in config/defaults.json. Consumers read from
it instead of hardcoding values:

- scripts/dev-safe.mjs: reads via JSON.parse(readFileSync(...))
- scripts/dev-with-automation.mjs: same
- scripts/check-sdk-version-sync.mjs: same (no longer regex-parses JS)
- docker/Dockerfile: config-gen build stage converts JSON to
  /opt/agent-canvas/defaults.env (shell-sourceable)
- docker/entrypoint.sh: sources defaults.env at startup; also adds
  session API key auto-generation so the image doesn't run wide-open
- .github/workflows/docker.yml: reads versions from JSON in a setup
  step (no more hardcoded env vars)

To bump a version, edit config/defaults.json only.

Co-authored-by: openhands <openhands@all-hands.dev>

* fix: address PR review feedback (#634)

- Fix PID tracking bug: move PIDS+=($!) inside if/elif branches so the
  else (automation-not-found) path doesn't add a stale PID
- chmod 600 session API key file to prevent credential leak
- Warn when using insecure default OH_SECRET_KEY in Docker entrypoint
- Add try/catch + field validation for config/defaults.json loading in
  check-sdk-version-sync.mjs
- Fix semver tag parsing: strip pre-release/build metadata, only create
  abbreviated tags (major.minor, major, latest) for stable releases
- Sanitize branch names for Docker tags (tr invalid chars, strip leading
  dot/dash) to handle branches with #, @, spaces, etc.
- Add arch validation before manifest merge (assert both amd64.json and
  arm64.json exist)
- Remove $schema reference to non-existent defaults.schema.json

Co-authored-by: openhands <openhands@all-hands.dev>

* fix: remove hardcoded version defaults from Dockerfile

Replace hardcoded ARG defaults (AGENT_SERVER_IMAGE, AUTOMATION_VERSION)
with empty ARGs. Values are always derived from config/defaults.json:
- CI: reads JSON in the workflow config step, passes --build-arg
- Local: new scripts/docker-build.mjs helper reads JSON and invokes
  docker build with the correct --build-arg values

Added npm run build:docker convenience script.

Co-authored-by: openhands <openhands@all-hands.dev>

* fix: stabilize snapshot tests and auto-generate Docker secret key

Two fixes:

1. **Flaky snapshot tests**: The 'Local pagination fixture' mock conversation
   used a fixed absolute timestamp (PAGINATION_BASE_TIME = May 13, 2026) for
   its created_at/updated_at, while 'Errored Project' used a relative
   timestamp (now - 7d). As real time progressed past the crossover point,
   their sort order in the sidebar flipped, causing 30/73 snapshot diffs on
   every PR. Fix: use relative timestamps (now - 6d) for the pagination
   fixture's conversation listing fields. The internal event timestamps
   (used by pagination tests) still use PAGINATION_BASE_TIME — only the
   sidebar ordering is affected.

2. **Docker OH_SECRET_KEY**: The entrypoint used a static insecure default
   for OH_SECRET_KEY and warned about it. Now mirrors the session API key
   pattern: auto-generate a cryptographic random key on first run, persist
   it to ~/.openhands/agent-canvas/secret-key.txt, and reuse on restart.
   Users can still override via the OH_SECRET_KEY env var. Removed the
   now-unused CONFIG_SECRET_KEY from the Docker defaults.env generation.
   Also deduped STATE_DIR computation (was repeated for session key path).

Co-authored-by: openhands <openhands@all-hands.dev>

* docs: update AGENTS.md with mock timestamp and Docker secret key notes

Co-authored-by: openhands <openhands@all-hands.dev>

* fix: include canvas_ui tool in Docker image

The Docker image was missing the tools/ directory and OH_EXTRA_PYTHON_PATH,
so the agent-server couldn't import canvas_ui_tool.py when the frontend
sent canvas_ui in the conversation tools list. This caused:

  HTTP 500: ToolDefinition 'canvas_ui' is not registered

Fix: COPY tools/ into the image and set OH_EXTRA_PYTHON_PATH in the
entrypoint, matching what scripts/dev-safe.mjs already does for local dev.

Co-authored-by: openhands <openhands@all-hands.dev>

---------

Co-authored-by: openhands <openhands@all-hands.dev>
This commit is contained in:
Rohit Malhotra
2026-05-20 00:24:45 -04:00
committed by GitHub
parent 1a6e629fda
commit 979e64fe19
12 changed files with 1070 additions and 84 deletions
+39
View File
@@ -0,0 +1,39 @@
# Dependencies
node_modules/
# Build outputs (rebuilt in Docker)
build/
dist/
# Development files
.git/
.github/
.openhands/
.agents/
.pr/
# Test files
__tests__/
tests/
__mocks__/
test-results/
playwright-report*/
coverage/
# IDE / OS
.vscode/
.idea/
*.swp
*.swo
.DS_Store
Thumbs.db
# Environment
.env
.env.*
!.env.sample
# Misc
artifacts/
*.tgz
*.log
+519
View File
@@ -0,0 +1,519 @@
---
name: Docker
on:
push:
branches: [main]
tags:
- "v*"
pull_request:
branches: [main]
workflow_dispatch:
inputs:
agent_server_image:
description: Agent Server base image (ghcr.io/openhands/agent-server:TAG)
type: string
default: ""
automation_version:
description: Automation server version (pip)
type: string
default: ""
image:
description: GHCR image name
type: string
default: ghcr.io/openhands/agent-canvas
# Cancel redundant runs for the same branch/PR.
concurrency:
group: ${{ github.workflow }}-${{ (github.head_ref && github.ref) || github.run_id }}
cancel-in-progress: true
permissions:
contents: read
packages: write
env:
IMAGE: ${{ inputs.image != '' && inputs.image || 'ghcr.io/openhands/agent-canvas' }}
# Use the PR head SHA for PR events so tags point at the actual code.
RELEVANT_SHA: ${{ github.event.pull_request.head.sha || github.sha }}
RELEVANT_REF: ${{ github.head_ref != '' && format('refs/heads/{0}', github.head_ref) || github.ref }}
jobs:
# ═══════════════════════════════════════════════════════════════════════════
# Build & Push (per-architecture, native runners)
# ═══════════════════════════════════════════════════════════════════════════
build-and-push-image:
name: Build & Push (${{ matrix.arch }})
# Skip fork PRs — they cannot authenticate to GHCR.
if: >
github.event_name == 'push' ||
github.event_name == 'workflow_dispatch' ||
(github.event_name == 'pull_request' &&
!github.event.pull_request.head.repo.fork)
strategy:
fail-fast: false
matrix:
include:
- arch: amd64
runner: ubuntu-24.04
platform: linux/amd64
- arch: arm64
runner: ubuntu-24.04-arm
platform: linux/arm64
runs-on: ${{ matrix.runner }}
timeout-minutes: 45
env:
ARCH: ${{ matrix.arch }}
PLATFORM: ${{ matrix.platform }}
steps:
- name: Checkout
uses: actions/checkout@v6
with:
ref: ${{ github.event.pull_request.head.sha || '' }}
- name: Set up Docker Buildx
uses: docker/setup-buildx-action@v4
- name: Log in to GHCR
uses: docker/login-action@v4
with:
registry: ghcr.io
username: ${{ github.actor }}
password: ${{ secrets.GITHUB_TOKEN }}
- name: Read defaults from config/defaults.json
id: config
run: |
# Single source of truth for version pins — no hardcoded values in this workflow.
AGENT_SERVER_VERSION=$(node -p "require('./config/defaults.json').versions.agentServer")
AGENT_SERVER_IMAGE_BASE=$(node -p "require('./config/defaults.json').images.agentServer")
AUTOMATION_VERSION=$(node -p "require('./config/defaults.json').versions.automation")
echo "agent_server_version=$AGENT_SERVER_VERSION" >> "$GITHUB_OUTPUT"
echo "default_agent_server_image=${AGENT_SERVER_IMAGE_BASE}:${AGENT_SERVER_VERSION}-python" >> "$GITHUB_OUTPUT"
echo "default_automation_version=$AUTOMATION_VERSION" >> "$GITHUB_OUTPUT"
- name: Compute metadata and tags
id: prep
run: |
SHORT_SHA=$(echo "$RELEVANT_SHA" | cut -c1-7)
echo "short_sha=$SHORT_SHA" >> "$GITHUB_OUTPUT"
# Resolve agent-server base image (input override > config/defaults.json)
if [ -n "${{ inputs.agent_server_image }}" ]; then
echo "agent_server_image=${{ inputs.agent_server_image }}" >> "$GITHUB_OUTPUT"
else
echo "agent_server_image=${{ steps.config.outputs.default_agent_server_image }}" >> "$GITHUB_OUTPUT"
fi
# Resolve automation version (input override > config/defaults.json)
if [ -n "${{ inputs.automation_version }}" ]; then
echo "automation_version=${{ inputs.automation_version }}" >> "$GITHUB_OUTPUT"
else
echo "automation_version=${{ steps.config.outputs.default_automation_version }}" >> "$GITHUB_OUTPUT"
fi
# Build arch-suffixed tags (e.g., sha-abc1234-amd64)
TAGS=""
add_tag() { TAGS="${TAGS:+${TAGS},}${IMAGE}:${1}-${ARCH}"; }
# SHA tags (always)
add_tag "sha-${SHORT_SHA}"
# Branch / PR / tag-based tags
if [[ "$RELEVANT_REF" == refs/heads/* ]]; then
# Sanitize branch name for Docker tag safety:
# replace any char outside [a-zA-Z0-9._-] with -, strip leading/trailing .-
BRANCH="${RELEVANT_REF#refs/heads/}"
BRANCH=$(echo "$BRANCH" | tr -c 'a-zA-Z0-9._-' '-' | sed 's/^[-.]//; s/[-.]*$//')
add_tag "$BRANCH"
fi
if [[ "${{ github.event_name }}" == "pull_request" ]]; then
add_tag "pr-${{ github.event.pull_request.number }}"
fi
if [[ "$RELEVANT_REF" == refs/tags/v* ]]; then
VERSION="${RELEVANT_REF#refs/tags/v}"
add_tag "$VERSION"
# Strip pre-release/build metadata for major.minor.patch derivation
VERSION_BASE="${VERSION%%-*}"
VERSION_BASE="${VERSION_BASE%%+*}"
# Only create abbreviated + latest tags for stable (non-pre-release) versions
if [[ "$VERSION" != *"-"* ]] && [[ "$VERSION" != *"+"* ]]; then
# major.minor
MINOR="${VERSION_BASE%.*}"
if [ "$MINOR" != "$VERSION_BASE" ]; then
add_tag "$MINOR"
fi
# major
MAJOR="${VERSION_BASE%%.*}"
if [ "$MAJOR" != "$VERSION_BASE" ] && [ "$MAJOR" != "$MINOR" ]; then
add_tag "$MAJOR"
fi
add_tag "latest"
fi
fi
echo "tags=$TAGS" >> "$GITHUB_OUTPUT"
echo "=== Build outputs ==="
echo "Short SHA: $SHORT_SHA"
echo "Tags: $TAGS"
echo "===================="
- name: Build & Push (${{ matrix.arch }})
id: build
uses: docker/build-push-action@v6
with:
context: .
file: docker/Dockerfile
platforms: ${{ matrix.platform }}
push: true
tags: ${{ steps.prep.outputs.tags }}
build-args: |
AGENT_SERVER_IMAGE=${{ steps.prep.outputs.agent_server_image }}
AUTOMATION_VERSION=${{ steps.prep.outputs.automation_version }}
OPENHANDS_BUILD_GIT_SHA=${{ env.RELEVANT_SHA }}
OPENHANDS_BUILD_GIT_REF=${{ env.RELEVANT_REF }}
cache-from: type=gha
cache-to: type=gha,mode=max
provenance: true
sbom: true
- name: Summary (${{ matrix.arch }})
run: |
echo "Image: ${{ env.IMAGE }}"
echo "Architecture: ${{ matrix.arch }}"
echo "Platform: ${{ matrix.platform }}"
echo "Short SHA: ${{ steps.prep.outputs.short_sha }}"
echo "Tags: ${{ steps.prep.outputs.tags }}"
echo "Build digest: ${{ steps.build.outputs.digest }}"
- name: Save build info for consolidation
run: |
mkdir -p build-info
jq -n \
--arg arch "${{ matrix.arch }}" \
--arg image "${{ env.IMAGE }}" \
--arg short_sha "${{ steps.prep.outputs.short_sha }}" \
--arg tags "${{ steps.prep.outputs.tags }}" \
--arg agent_server_image "${{ steps.prep.outputs.agent_server_image }}" \
--arg automation_version "${{ steps.prep.outputs.automation_version }}" \
--arg platform "${{ matrix.platform }}" \
--arg git_sha "${{ env.RELEVANT_SHA }}" \
--arg git_ref "${{ env.RELEVANT_REF }}" \
'{arch: $arch, image: $image, short_sha: $short_sha, tags: $tags, agent_server_image: $agent_server_image, automation_version: $automation_version, platform: $platform, git_sha: $git_sha, git_ref: $git_ref}' \
> "build-info/${{ matrix.arch }}.json"
cat "build-info/${{ matrix.arch }}.json"
- name: Upload build info artifact
uses: actions/upload-artifact@v7
with:
name: build-info-${{ matrix.arch }}
path: build-info/${{ matrix.arch }}.json
retention-days: 1
# ═══════════════════════════════════════════════════════════════════════════
# Merge Multi-Arch Manifests
# ═══════════════════════════════════════════════════════════════════════════
merge-manifests:
name: Merge Multi-Arch Manifests
needs: build-and-push-image
if: >
github.event_name == 'push' ||
github.event_name == 'workflow_dispatch' ||
(github.event_name == 'pull_request' &&
!github.event.pull_request.head.repo.fork)
runs-on: ubuntu-24.04
steps:
- name: Download build info artifacts
uses: actions/download-artifact@v8
with:
pattern: build-info-*
merge-multiple: true
path: build-info
- name: Extract SHORT_SHA from build info
id: get_sha
run: |
SHORT_SHA=$(jq -r '.short_sha' build-info/amd64.json)
echo "short_sha=$SHORT_SHA" >> "$GITHUB_OUTPUT"
echo "Using SHORT_SHA: $SHORT_SHA"
- name: Set up Docker Buildx
uses: docker/setup-buildx-action@v4
- name: Log in to GHCR
uses: docker/login-action@v4
with:
registry: ghcr.io
username: ${{ github.actor }}
password: ${{ secrets.GITHUB_TOKEN }}
- name: Create and push multi-arch manifests
id: create_manifests
run: |
# Validate both architectures built successfully
if [[ ! -f build-info/amd64.json ]] || [[ ! -f build-info/arm64.json ]]; then
echo "::error::Missing architecture builds (need both amd64.json and arm64.json)"
echo "Available: $(ls -1 build-info/*.json 2>/dev/null || echo 'none')"
exit 1
fi
SHORT_SHA=${{ steps.get_sha.outputs.short_sha }}
AMD64_TAGS_CSV=$(jq -r '.tags' build-info/amd64.json)
declare -A SEEN_MANIFEST_TAGS=()
MANIFEST_TAGS=()
create_manifest() {
local manifest_tag=$1
local source_tag=${2:-$1}
echo "Creating multi-arch manifest: ${IMAGE}:${manifest_tag}"
docker buildx imagetools create -t "${IMAGE}:${manifest_tag}" \
"${IMAGE}:${source_tag}-amd64" \
"${IMAGE}:${source_tag}-arm64"
echo "Inspecting multi-arch manifest:"
docker buildx imagetools inspect "${IMAGE}:${manifest_tag}"
echo "✓ Multi-arch manifest created: ${IMAGE}:${manifest_tag}"
}
IFS=',' read -ra AMD64_TAGS <<< "$AMD64_TAGS_CSV"
for AMD64_IMAGE_TAG in "${AMD64_TAGS[@]}"; do
if [ -z "$AMD64_IMAGE_TAG" ]; then
continue
fi
TAG_NAME=${AMD64_IMAGE_TAG#${IMAGE}:}
if [ "$TAG_NAME" = "$AMD64_IMAGE_TAG" ] || [[ ! "$TAG_NAME" == *-amd64 ]]; then
echo "Skipping unexpected architecture tag: $AMD64_IMAGE_TAG"
continue
fi
MANIFEST_TAG=${TAG_NAME%-amd64}
if [ -n "${SEEN_MANIFEST_TAGS[$MANIFEST_TAG]+x}" ]; then
continue
fi
SEEN_MANIFEST_TAGS[$MANIFEST_TAG]=1
MANIFEST_TAGS+=("$MANIFEST_TAG")
create_manifest "$MANIFEST_TAG"
done
# Preserve a latest alias on main pushes.
if [ "${{ github.ref }}" == "refs/heads/main" ]; then
LATEST_TAG="latest"
create_manifest "$LATEST_TAG" "main"
MANIFEST_TAGS+=("$LATEST_TAG")
fi
MANIFEST_TAG_CSV=$(IFS=,; echo "${MANIFEST_TAGS[*]}")
echo "manifest_tags=$MANIFEST_TAG_CSV" >> "$GITHUB_OUTPUT"
# Save manifest info for consolidation
mkdir -p manifest-info
jq -n \
--arg image "${{ env.IMAGE }}" \
--arg short_sha "$SHORT_SHA" \
--arg manifest_tags "$MANIFEST_TAG_CSV" \
'{image: $image, short_sha: $short_sha, manifest_tags: $manifest_tags}' \
> manifest-info/manifests.json
cat manifest-info/manifests.json
- name: Upload manifest info artifact
uses: actions/upload-artifact@v7
with:
name: manifest-info
path: manifest-info/manifests.json
retention-days: 1
# ═══════════════════════════════════════════════════════════════════════════
# Consolidate Build Information
# ═══════════════════════════════════════════════════════════════════════════
consolidate-build-info:
name: Consolidate Build Information
needs: [build-and-push-image, merge-manifests]
if: github.event_name == 'pull_request' && always() && (needs.build-and-push-image.result == 'success' || needs.build-and-push-image.result == 'failure')
runs-on: ubuntu-24.04
outputs:
build_summary: ${{ steps.consolidate.outputs.build_summary }}
steps:
- name: Download build info artifacts
uses: actions/download-artifact@v8
with:
pattern: build-info-*
merge-multiple: true
path: build-info
- name: Download manifest info artifacts
uses: actions/download-artifact@v8
with:
name: manifest-info
path: manifest-info
continue-on-error: true
- name: Consolidate build information from artifacts
id: consolidate
run: |
echo "Processing build info artifacts..."
ls -la build-info/
IMAGE=""
SHORT_SHA=""
ALL_TAGS=""
AGENT_SERVER_IMAGE=""
AUTOMATION_VERSION=""
GIT_SHA=""
GIT_REF=""
ARCHS=""
for info_file in build-info/*.json; do
if [[ ! -f "$info_file" ]]; then
continue
fi
echo "=== Processing $info_file ==="
cat "$info_file"
ARCH=$(jq -r '.arch' "$info_file")
FILE_IMAGE=$(jq -r '.image' "$info_file")
FILE_SHA=$(jq -r '.short_sha' "$info_file")
FILE_TAGS=$(jq -r '.tags' "$info_file")
if [[ -z "$IMAGE" ]]; then
IMAGE="$FILE_IMAGE"
SHORT_SHA="$FILE_SHA"
AGENT_SERVER_IMAGE=$(jq -r '.agent_server_image' "$info_file")
AUTOMATION_VERSION=$(jq -r '.automation_version' "$info_file")
GIT_SHA=$(jq -r '.git_sha' "$info_file")
GIT_REF=$(jq -r '.git_ref' "$info_file")
fi
ARCHS="${ARCHS:+${ARCHS}, }${ARCH}"
if [[ -n "$FILE_TAGS" ]]; then
TAG_LIST=$(echo "$FILE_TAGS" | tr ',' '\n')
ALL_TAGS="${ALL_TAGS:+${ALL_TAGS}
}${TAG_LIST}"
fi
done
# Add manifest tags
if [[ -f "manifest-info/manifests.json" ]]; then
MANIFEST_TAG_CSV=$(jq -r '.manifest_tags' manifest-info/manifests.json)
MANIFEST_TAG_LIST=$(echo "$MANIFEST_TAG_CSV" | tr ',' '\n' | sed "s|^|${IMAGE}:|")
ALL_TAGS="${ALL_TAGS:+${ALL_TAGS}
}${MANIFEST_TAG_LIST}"
fi
BUILD_SUMMARY=$(jq -n \
--arg image "$IMAGE" \
--arg short_sha "$SHORT_SHA" \
--arg all_tags "$ALL_TAGS" \
--arg archs "$ARCHS" \
--arg agent_server_image "$AGENT_SERVER_IMAGE" \
--arg automation_version "$AUTOMATION_VERSION" \
--arg git_sha "$GIT_SHA" \
--arg git_ref "$GIT_REF" \
--arg ghcr_url "https://github.com/OpenHands/agent-canvas/pkgs/container/agent-canvas" \
'{image: $image, short_sha: $short_sha, all_tags: $all_tags, architectures: $archs, agent_server_image: $agent_server_image, automation_version: $automation_version, git_sha: $git_sha, git_ref: $git_ref, ghcr_package_url: $ghcr_url}')
echo "Consolidated build summary:"
echo "$BUILD_SUMMARY" | jq .
{
echo 'build_summary<<EOF'
echo "$BUILD_SUMMARY"
echo 'EOF'
} >> "$GITHUB_OUTPUT"
# ═══════════════════════════════════════════════════════════════════════════
# Update PR description with image info (PRs only)
# ═══════════════════════════════════════════════════════════════════════════
update-pr-description:
name: Update PR description with Docker image
needs: consolidate-build-info
if: github.event_name == 'pull_request' && needs.consolidate-build-info.result == 'success'
runs-on: ubuntu-24.04
permissions:
contents: read
pull-requests: write
steps:
- name: Generate PR description from build summary
id: generate_description
run: |
BUILD_SUMMARY='${{ needs.consolidate-build-info.outputs.build_summary }}'
echo "Build summary received:"
echo "$BUILD_SUMMARY" | jq .
IMAGE=$(echo "$BUILD_SUMMARY" | jq -r '.image')
SHORT_SHA=$(echo "$BUILD_SUMMARY" | jq -r '.short_sha')
GHCR_URL=$(echo "$BUILD_SUMMARY" | jq -r '.ghcr_package_url')
ALL_TAGS=$(echo "$BUILD_SUMMARY" | jq -r '.all_tags')
ARCHS=$(echo "$BUILD_SUMMARY" | jq -r '.architectures')
AGENT_SERVER_IMAGE=$(echo "$BUILD_SUMMARY" | jq -r '.agent_server_image')
AUTOMATION_VERSION=$(echo "$BUILD_SUMMARY" | jq -r '.automation_version')
GIT_SHA=$(echo "$BUILD_SUMMARY" | jq -r '.git_sha')
PR_CONTENT=$(cat << EOF
<!-- AGENT_CANVAS_DOCKER_START -->
---
**🐳 Docker images for this PR**
• **GHCR package:** ${GHCR_URL}
| Component | Value |
|---|---|
| **Image** | \`${IMAGE}\` |
| **Architectures** | ${ARCHS} |
| **Agent Server** | \`${AGENT_SERVER_IMAGE}\` |
| **Automation** | \`openhands-automation==${AUTOMATION_VERSION}\` |
| **Commit** | \`${GIT_SHA}\` |
**Pull (multi-arch manifest)**
\`\`\`bash
# Multi-arch manifest — Docker automatically pulls the correct architecture
docker pull ${IMAGE}:sha-${SHORT_SHA}
\`\`\`
**Run**
\`\`\`bash
docker run -it --rm \\
-p 8000:8000 \\
${IMAGE}:sha-${SHORT_SHA}
\`\`\`
**All tags pushed for this build**
\`\`\`
${ALL_TAGS}
\`\`\`
**About Multi-Architecture Support**
- Each tag (e.g., \`sha-${SHORT_SHA}\`) is a **multi-arch manifest** supporting both **amd64** and **arm64**
- Docker automatically pulls the correct architecture for your platform
- Individual architecture tags (e.g., \`sha-${SHORT_SHA}-amd64\`) are also available if needed
<!-- AGENT_CANVAS_DOCKER_END -->
EOF
)
{
echo 'pr_content<<EOF'
echo "$PR_CONTENT"
echo 'EOF'
} >> "$GITHUB_OUTPUT"
- name: Update PR description with docker image details
uses: nefrob/pr-description@v1.2.0
with:
content: ${{ steps.generate_description.outputs.pr_content }}
regex: "<!-- AGENT_CANVAS_DOCKER_START -->.*?<!-- AGENT_CANVAS_DOCKER_END -->"
regexFlags: s
token: ${{ secrets.GITHUB_TOKEN }}
+10 -1
View File
@@ -119,6 +119,7 @@ you are running inside of — NOT the automation backend.
- Snapshots are organized by `{snapshotDir}/{testFilePath}/{projectName}/{arg}.png` (configured in `playwright.config.ts`).
- **Conversation page snapshot tests**: The dev server uses MSW service workers for API mocking. For conversation-page tests, rely on MSW's pre-defined mock conversations (IDs "1", "2", "3" in `src/mocks/conversation-handlers.ts`) rather than fighting Playwright route interception. MSW's service worker intercepts requests before Playwright `page.route()` can; Playwright route interceptors only see requests that escape the service worker. Stub WebSocket via `page.addInitScript()` and inject events into the Zustand store via the exposed `window.__OH_EVENT_STORE__` API. Use `test.describe.configure({ mode: "serial" })` for conversation tests since the WebSocket stub + heavier page setup can cause intermittent failures in parallel mode.
- **Baseline generation for CI**: Baselines generated locally will NOT match CI (different OS, fonts, rendering). Baselines are regenerated automatically on every push to `main`. After adding new snapshot tests, open a PR — the new snapshots will be shown as "🆕 New" in the PR comment and become the baseline when the PR merges. To force-refresh baselines from main without waiting for a code push, trigger the "Snapshot Tests" workflow manually with `force_update=true`.
- **Mock conversation timestamps must ALL be `now`-relative**: `src/mocks/conversation-handlers.ts` defines mock conversations sorted by `updated_at` descending in the sidebar. Every conversation's `created_at`/`updated_at` must use `now - X * days` (relative to module-load time), never a fixed absolute date like `PAGINATION_BASE_TIME`. Mixing the two strategies causes a sort-order crossover as real time passes — the fixed-date conversation ages past a relative one and they swap positions, breaking every snapshot that includes the sidebar. `PAGINATION_BASE_TIME` is kept only for internal event timestamps used by pagination tests; conversation listing timestamps are decoupled from it.
- **MSW handler state is PAGE-level JS, not service-worker state**: In MSW 2.x browser mode the request handlers (including mutable Maps like `automations`) are compiled into the client bundle and run in the main thread. `page.reload()` re-initialises all module-level state (e.g. `const automations = new Map(...)` runs fresh on every page load). Tests that need to show an "empty list" state must NOT call `page.reload()` after deleting items. Instead: make the DELETE fetches from `page.evaluate()` (which DO go through MSW), then call `window.__TEST_INVALIDATE_QUERIES__()` (exposed in mock mode by `entry.client.tsx`) to trigger React Query's cache invalidation in-place without a reload. Example: the automations empty-state snapshot test in `tests/e2e/snapshots/automations.snapshot.spec.ts`.
## Live End-to-End Test Framework
@@ -337,7 +338,7 @@ return new ConversationClient(getAgentServerClientOptions()).someMethod(...);
- `/api/*`, `/sockets`, etc. → agent server (:18000)
- `/*` (default) → frontend server (:3001), either Vite or static depending on launcher mode
- Environment variables: `PORT` (ingress port, default: 8000), `OH_AUTOMATION_GIT_REF` (git ref, overrides default version), `OH_AUTOMATION_VERSION` (default: `1.0.0a3`), `AUTOMATION_LOCAL_API_KEY` (optional, use a fixed key; default: persisted generated key), `OH_AUTOMATION_API_KEY_PATH` (override the persisted default key path)
- `scripts/check-sdk-version-sync.mjs` checks the released `openhands-automation` package against `DEFAULT_AUTOMATION_SDK_VERSION` in `scripts/dev-with-automation.mjs`; that value may intentionally lag `DEFAULT_AGENT_SERVER_VERSION` while automation has not yet published a matching release.
- `scripts/check-sdk-version-sync.mjs` checks the released `openhands-automation` package against `versions.automationSdk` in `config/defaults.json`; that value may intentionally lag `versions.agentServer` while automation has not yet published a matching release.
- Access points: `http://localhost:8000/` (main UI), `http://localhost:8000/api/automation/docs` (API docs)
- Security: `AUTOMATION_LOCAL_API_KEY` defaults to a generated key persisted across restarts because static frontend builds bake it into `VITE_AUTOMATION_API_KEY`. Set the env var explicitly to rotate or pin it. The cipher key (`OH_SECRET_KEY`) keeps a static default for local dev since it's used for encrypting/decrypting persisted settings values.
- `scripts/ingress.mjs` is a standalone HTTP reverse proxy that can be used independently to route traffic to multiple backends based on URL path prefix.
@@ -438,4 +439,12 @@ return new ConversationClient(getAgentServerClientOptions()).someMethod(...);
- ESLint config (flat, ESLint 9): the project uses `eslint.config.js` (not `.eslintrc`) and runs on `eslint@9.x`, not 10. The constraint pinning us below 10 is `eslint-plugin-react@7.37.x`, which still calls `context.getFilename()` at rule-load time — that API was removed in ESLint 10 and `@eslint/compat`'s `fixupPluginRules` does NOT shim it. Don't try to bump eslint past 9 until eslint-plugin-react ships a v10-compatible release. Import rules come from `eslint-plugin-import-x` (the maintained fork of `eslint-plugin-import`) but are registered under both `import-x/` and `import/` prefixes via `plugins: { import: importXPlugin, ... }` so existing `// eslint-disable-next-line import/...` directives keep working. `linterOptions.reportUnusedDisableDirectives` is set to `"warn"` (not "off") so stale airbnb-era disable comments still surface in lint output without failing CI. The TS-overrides block has an `ignores: ["src/hooks/query/query-keys.ts"]` so the `no-restricted-syntax` rule banning raw `["settings", ...]` query keys doesn't fire on the file that defines the helpers themselves. No `.npmrc` / `legacy-peer-deps` flag is needed — all our plugins declare ESLint 9 peer compatibility.
- **Centralized config**: `config/defaults.json` is the single source of truth for version pins (agent-server, automation, automation SDK), port defaults, persistence paths, package names, and the dev secret key. All consumers read from this file:
- JS scripts (`dev-safe.mjs`, `dev-with-automation.mjs`, `check-sdk-version-sync.mjs`) read it via `JSON.parse(readFileSync(...))`.
- Docker: a `config-gen` build stage converts the JSON to `/opt/agent-canvas/defaults.env` (shell-sourceable); `entrypoint.sh` sources it at startup.
- CI workflow: a `Read defaults from config/defaults.json` step uses `node -p` to extract values into `$GITHUB_OUTPUT`.
- Dockerfile ARG defaults are kept as fallbacks for local `docker build` without the CI workflow; CI always passes `--build-arg` overrides from the JSON.
- To bump a version, edit `config/defaults.json` only — the JS scripts, Docker build, and CI workflow all derive their values from it.
- Docker all-in-one image: `.github/workflows/docker.yml` builds and publishes `ghcr.io/openhands/agent-canvas` — a combined image that bundles the agent-server (from `ghcr.io/openhands/agent-server`), the automation server (`openhands-automation` via pip), and the agent-canvas frontend (static build). The Dockerfile lives at `docker/Dockerfile`, the entrypoint at `docker/entrypoint.sh`. The workflow structure mirrors the SDK repo's `server.yml`: a `build-and-push-image` matrix job (2 × arch: amd64 on `ubuntu-24.04`, arm64 on `ubuntu-24.04-arm`) pushes arch-suffixed tags, then `merge-manifests` creates multi-arch manifests via `docker buildx imagetools create`, then `consolidate-build-info` aggregates artifacts, and `update-pr-description` updates the PR body (using `<!-- AGENT_CANVAS_DOCKER_START -->` / `<!-- AGENT_CANVAS_DOCKER_END -->` markers). The workflow triggers on push to main, `v*` tags (releases), PRs, and `workflow_dispatch`. On release tags it also pushes semver tags (e.g. `1.2.3`, `1.2`, `1`, `latest`). Fork PRs are skipped (no GHCR auth). The image exposes port 8000 as a unified entry point: `/api/automation/*` → automation (:18001), `/api/*` → agent-server (:18000), `/*` → static frontend. The entrypoint auto-generates **both** the session API key and `OH_SECRET_KEY` (persisted to `~/.openhands/agent-canvas/session-api-key.txt` and `secret-key.txt` respectively) when none is provided, so the image runs secure by default. Users can override either via env var (`OH_SECRET_KEY`, `SESSION_API_KEY` / `OH_SESSION_API_KEYS_0`). Unlike `scripts/dev-safe.mjs` (which uses a static default secret key for local dev convenience), the Docker entrypoint never falls back to a known default.
- Cloud conversation resume gating: when a cloud conversation is closed from the UI (`pauseCloudSandbox` is called), the conversation's `conversation_url` is NOT cleared -- it still points to the old sandbox host. `WebSocketProviderWrapper` must suppress the URL (pass `null` to `ConversationWebSocketProvider`) while `sandbox_status === "PAUSED"`, otherwise the WebSocket immediately tries the stale URL before the sandbox wakes. Symmetrically, `useActiveConversation`'s refetch interval must fast-poll (3 s) on both `!conversation_url` AND `sandbox_status === "PAUSED"` -- checking only the missing URL would leave the hook on the 30 s interval while the sandbox is resuming. The resume sequence: navigate -> sandbox PAUSED detected -> `resumeCloudSandbox` called (in `conversation.tsx`) -> fast-poll detects RUNNING -> `conversationUrl` unblocked -> WebSocket connects.
+38
View File
@@ -0,0 +1,38 @@
{
"_comment": "Single source of truth for version pins, ports, paths, and defaults shared across the npm and Docker install paths. Read by scripts/dev-safe.mjs, scripts/dev-with-automation.mjs, docker/entrypoint.sh (via generated defaults.env), and .github/workflows/docker.yml.",
"versions": {
"agentServer": "1.22.1",
"automation": "1.0.0a3",
"automationSdk": "1.22.1"
},
"images": {
"agentServer": "ghcr.io/openhands/agent-server",
"agentCanvas": "ghcr.io/openhands/agent-canvas"
},
"ports": {
"agentServer": 18000,
"automation": 18001,
"proxy": 8000
},
"paths": {
"stateSubdir": "agent-canvas",
"conversations": "agent-canvas/conversations",
"bashEvents": "agent-canvas/bash_events",
"automationDb": "automation/automations.db"
},
"packages": {
"agentServer": "openhands-agent-server",
"automation": "openhands-automation",
"tools": "openhands-tools",
"workspace": "openhands-workspace"
},
"defaults": {
"secretKey": "openhands-dev-secret-key-change-in-prod"
}
}
+131
View File
@@ -0,0 +1,131 @@
# syntax=docker/dockerfile:1.7
# ═══════════════════════════════════════════════════════════════════════════════
# agent-canvas all-in-one Docker image
#
# Combines three services into a single image:
# 1. Agent Server — from ghcr.io/openhands/agent-server (upstream SDK image)
# 2. Automation — installed via pip from openhands-automation
# 3. Frontend — agent-canvas static build served by Node.js
#
# The entrypoint starts all three services and an ingress proxy that unifies
# them behind a single port (default 8000):
# /api/automation/* → automation backend (:18001)
# /api/*, /sockets → agent server (:18000)
# /* (default) → static frontend + SPA fallback
# ═══════════════════════════════════════════════════════════════════════════════
# ── Build args ────────────────────────────────────────────────────────────────
# No hardcoded defaults — values are derived from config/defaults.json.
# CI passes these via --build-arg; for local builds use:
# node scripts/docker-build.mjs (recommended, reads JSON for you)
# docker build --build-arg AGENT_SERVER_IMAGE=... --build-arg AUTOMATION_VERSION=... -f docker/Dockerfile .
ARG AGENT_SERVER_IMAGE
ARG AUTOMATION_VERSION
# ── Stage 1: Build frontend ──────────────────────────────────────────────────
FROM node:24-slim AS frontend-build
WORKDIR /build
# Cache-friendly: package files first
COPY package.json package-lock.json ./
RUN npm ci
# Copy everything needed for the build
COPY . .
# Build the static frontend
RUN npm run build
# ── Stage 1b: Generate shell-sourceable defaults from config/defaults.json ──
# This avoids needing jq/python at container runtime to parse the JSON.
FROM node:24-slim AS config-gen
COPY config/defaults.json /tmp/defaults.json
RUN node -e " \
const c = JSON.parse(require('fs').readFileSync('/tmp/defaults.json','utf-8')); \
const lines = [ \
'CONFIG_AGENT_SERVER_PORT=' + c.ports.agentServer, \
'CONFIG_AUTOMATION_PORT=' + c.ports.automation, \
'CONFIG_PROXY_PORT=' + c.ports.proxy, \
'CONFIG_STATE_SUBDIR=' + c.paths.stateSubdir, \
'CONFIG_CONVERSATIONS=' + c.paths.conversations, \
'CONFIG_BASH_EVENTS=' + c.paths.bashEvents, \
'CONFIG_AUTOMATION_DB=' + c.paths.automationDb, \
]; \
require('fs').writeFileSync('/tmp/defaults.env', lines.join('\n') + '\n'); \
"
# ── Stage 2: Combined image ──────────────────────────────────────────────────
FROM ${AGENT_SERVER_IMAGE} AS final
ARG AUTOMATION_VERSION
ARG OPENHANDS_BUILD_GIT_SHA=unknown
ARG OPENHANDS_BUILD_GIT_REF=unknown
LABEL org.opencontainers.image.title="agent-canvas"
LABEL org.opencontainers.image.description="All-in-one agent-canvas: Agent Server + Automation + Frontend"
LABEL org.opencontainers.image.source="https://github.com/OpenHands/agent-canvas"
LABEL org.opencontainers.image.revision="${OPENHANDS_BUILD_GIT_SHA}"
ENV AGENT_CANVAS_BUILD_GIT_SHA=${OPENHANDS_BUILD_GIT_SHA}
ENV AGENT_CANVAS_BUILD_GIT_REF=${OPENHANDS_BUILD_GIT_REF}
USER root
# Install system deps required by automation's transitive dependencies
# (asyncpg needs libpq, which the agent-server base image may not include).
RUN if command -v apt-get >/dev/null 2>&1; then \
apt-get update && \
apt-get install -y --no-install-recommends libpq-dev && \
rm -rf /var/lib/apt/lists/*; \
fi
# Install automation server via pip.
# Shared deps (openhands-sdk, fastapi, uvicorn, pydantic, httpx, …) are
# already satisfied by the agent-server base image, so only automation-
# specific packages (asyncpg, sqlalchemy, boto3, gcloud, …) are added.
RUN uv pip install --system "openhands-automation==${AUTOMATION_VERSION}" 2>/dev/null \
|| pip install --no-cache-dir "openhands-automation==${AUTOMATION_VERSION}"
# Copy the frontend build output.
# react-router.config.ts unpacks build/client/ into build/ for non-Vercel builds.
COPY --from=frontend-build /build/build /opt/agent-canvas/frontend
# Copy the static-server script (serves frontend + proxies to backends)
COPY scripts/static-server.mjs /opt/agent-canvas/static-server.mjs
# Copy custom tools (e.g. canvas_ui_tool.py) so the agent-server can import
# them via tool_module_qualnames. OH_EXTRA_PYTHON_PATH is set in entrypoint.sh.
COPY tools/ /opt/agent-canvas/tools/
# Copy generated defaults.env (from config/defaults.json via config-gen stage)
COPY --from=config-gen /tmp/defaults.env /opt/agent-canvas/defaults.env
# Copy the entrypoint
COPY docker/entrypoint.sh /opt/agent-canvas/entrypoint.sh
RUN chmod +x /opt/agent-canvas/entrypoint.sh
# Pre-create persistence directories with correct ownership so the
# openhands user can write to them even when Docker creates anonymous
# volumes (which default to root).
RUN mkdir -p /home/openhands/.openhands/agent-canvas/conversations \
/home/openhands/.openhands/agent-canvas/bash_events \
/home/openhands/.openhands/automation \
/projects && \
chown -R openhands:openhands /home/openhands/.openhands /projects
USER openhands
# Persistence volumes:
# /home/openhands/.openhands — settings, secrets, conversations, automation DB
# /projects — user code the agent can read/edit
# Bind-mount these for data to survive container restarts:
# docker run -v ~/.openhands:/home/openhands/.openhands -v ~/projects:/projects ...
VOLUME ["/home/openhands/.openhands", "/projects"]
# The entrypoint starts all services and the ingress proxy.
# Port 8000 is the unified entry point.
EXPOSE 8000
ENTRYPOINT ["tini", "--", "/opt/agent-canvas/entrypoint.sh"]
+191
View File
@@ -0,0 +1,191 @@
#!/usr/bin/env bash
# ═══════════════════════════════════════════════════════════════════════════════
# agent-canvas all-in-one entrypoint
#
# Starts three services:
# 1. Agent Server on port $AGENT_SERVER_PORT (default 18000)
# 2. Automation on port $AUTOMATION_PORT (default 18001)
# 3. Static server on port $PORT (default 8000)
# Routes /api/automation/* → automation, /api/* → agent-server,
# and serves the frontend static build for everything else.
#
# Environment variables:
# PORT – Unified entry point port (default: 8000)
# AGENT_SERVER_PORT – Internal agent-server port (default: 18000)
# AUTOMATION_PORT – Internal automation port (default: 18001)
# OH_SECRET_KEY – Secret key for settings encryption (auto-generated
# and persisted if not provided)
# OPENHANDS_AUTOMATION_API_KEY – API key for automation backend auth
# Any agent-server or automation env vars are passed through.
# ═══════════════════════════════════════════════════════════════════════════════
set -uo pipefail
log() { printf '[agent-canvas] %s\n' "$*"; }
log_error() { printf '[agent-canvas] ERROR: %s\n' "$*" >&2; }
# ── Load centralized defaults (generated from config/defaults.json at build) ─
# shellcheck source=/dev/null
if [ -f /opt/agent-canvas/defaults.env ]; then
# shellcheck disable=SC1091
. /opt/agent-canvas/defaults.env
fi
PORT="${PORT:-${CONFIG_PROXY_PORT:-8000}}"
AGENT_SERVER_PORT="${AGENT_SERVER_PORT:-${CONFIG_AGENT_SERVER_PORT:-18000}}"
AUTOMATION_PORT="${AUTOMATION_PORT:-${CONFIG_AUTOMATION_PORT:-18001}}"
# Persistence paths — keep settings, conversations, bash history under a
# single well-known directory that the VOLUME directive exposes.
OPENHANDS_DIR="${HOME}/.openhands"
STATE_DIR="${OPENHANDS_DIR}/${CONFIG_STATE_SUBDIR:-agent-canvas}"
export OH_PERSISTENCE_DIR="${OH_PERSISTENCE_DIR:-${OPENHANDS_DIR}}"
export OH_CONVERSATIONS_PATH="${OH_CONVERSATIONS_PATH:-${OPENHANDS_DIR}/${CONFIG_CONVERSATIONS:-agent-canvas/conversations}}"
export OH_BASH_EVENTS_DIR="${OH_BASH_EVENTS_DIR:-${OPENHANDS_DIR}/${CONFIG_BASH_EVENTS:-agent-canvas/bash_events}}"
# OH_SECRET_KEY is required for settings/secrets encryption. Without it the
# agent-server refuses to return encrypted secrets → conversation creation
# fails with a 503. Auto-generate and persist (just like the session API key)
# so the image never runs with a known default.
SECRET_KEY_FILE="${STATE_DIR}/secret-key.txt"
if [ -z "${OH_SECRET_KEY:-}" ]; then
if [ -f "$SECRET_KEY_FILE" ]; then
OH_SECRET_KEY="$(cat "$SECRET_KEY_FILE")"
else
OH_SECRET_KEY="$(head -c 32 /dev/urandom | od -An -tx1 | tr -d ' \n')"
mkdir -p "$(dirname "$SECRET_KEY_FILE")"
printf '%s' "$OH_SECRET_KEY" > "$SECRET_KEY_FILE"
chmod 600 "$SECRET_KEY_FILE"
log "Generated OH_SECRET_KEY (persisted to $SECRET_KEY_FILE)"
fi
fi
export OH_SECRET_KEY
# Session API key — generate one if not provided so the image doesn't run
# wide-open by default. Persisted so restarts reuse the same key.
SESSION_KEY_FILE="${STATE_DIR}/session-api-key.txt"
if [ -z "${OH_SESSION_API_KEYS_0:-}" ] && [ -z "${SESSION_API_KEY:-}" ]; then
if [ -f "$SESSION_KEY_FILE" ]; then
SESSION_API_KEY="$(cat "$SESSION_KEY_FILE")"
else
SESSION_API_KEY="$(head -c 32 /dev/urandom | od -An -tx1 | tr -d ' \n')"
mkdir -p "$(dirname "$SESSION_KEY_FILE")"
printf '%s' "$SESSION_API_KEY" > "$SESSION_KEY_FILE"
chmod 600 "$SESSION_KEY_FILE"
log "Generated session API key (persisted to $SESSION_KEY_FILE)"
fi
export OH_SESSION_API_KEYS_0="$SESSION_API_KEY"
fi
# AGENT_SERVER_URL — needed by automation sandbox callbacks.
export AGENT_SERVER_URL="${AGENT_SERVER_URL:-http://127.0.0.1:${AGENT_SERVER_PORT}}"
# Make custom tools (e.g. canvas_ui_tool.py) importable by the agent-server
# via tool_module_qualnames. Matches what scripts/dev-safe.mjs does with
# OH_EXTRA_PYTHON_PATH: config.canvasToolsDir.
export OH_EXTRA_PYTHON_PATH="${OH_EXTRA_PYTHON_PATH:-/opt/agent-canvas/tools}"
# Track child PIDs so we can clean up on exit.
PIDS=()
cleanup() {
log "Shutting down..."
for pid in "${PIDS[@]}"; do
kill "$pid" 2>/dev/null || true
done
wait 2>/dev/null || true
}
trap cleanup EXIT SIGINT SIGTERM
# ── 1. Start Agent Server ────────────────────────────────────────────────────
log "Starting agent-server on port $AGENT_SERVER_PORT..."
if command -v openhands-agent-server >/dev/null 2>&1; then
# Binary build (production image)
openhands-agent-server --port "$AGENT_SERVER_PORT" &
elif [ -x /agent-server/.venv/bin/python ]; then
# Source build (development image)
/agent-server/.venv/bin/python -m openhands.agent_server --port "$AGENT_SERVER_PORT" &
else
log_error "Cannot find agent-server binary or source venv."
exit 1
fi
PIDS+=($!)
# ── 2. Start Automation Server ───────────────────────────────────────────────
log "Starting automation server on port $AUTOMATION_PORT..."
# Disable the automation's own frontend — agent-canvas provides the UI.
export AUTOMATION_FRONTEND_DIR=""
# Default to SQLite so the automation server works out of the box without
# an external PostgreSQL instance. Users can override AUTOMATION_DB_URL to
# point at a real Postgres for production deployments.
if [ -z "${AUTOMATION_DB_URL:-}" ]; then
AUTOMATION_DB_FILE="${OPENHANDS_DIR}/${CONFIG_AUTOMATION_DB:-automation/automations.db}"
mkdir -p "$(dirname "$AUTOMATION_DB_FILE")"
export AUTOMATION_DB_URL="sqlite+aiosqlite:///${AUTOMATION_DB_FILE}"
log "Using SQLite database: $AUTOMATION_DB_URL"
fi
# The automation server uses uvicorn. Set AUTOMATION_PORT via its CLI.
if command -v uvicorn >/dev/null 2>&1; then
uvicorn openhands.automation.app:app \
--host 0.0.0.0 \
--port "$AUTOMATION_PORT" &
PIDS+=($!)
elif python -c "import openhands.automation" 2>/dev/null; then
python -m uvicorn openhands.automation.app:app \
--host 0.0.0.0 \
--port "$AUTOMATION_PORT" &
PIDS+=($!)
else
log "WARNING: Automation server not found, skipping."
fi
# ── 3. Wait for backends to be ready ─────────────────────────────────────────
wait_for_port() {
local port=$1 name=$2 max_wait=${3:-30}
local elapsed=0
while ! (echo >/dev/tcp/127.0.0.1/"$port") 2>/dev/null; do
sleep 1
elapsed=$((elapsed + 1))
if [ "$elapsed" -ge "$max_wait" ]; then
log "WARNING: $name on port $port did not become ready within ${max_wait}s"
return 1
fi
done
log "$name is ready on port $port"
}
wait_for_port "$AGENT_SERVER_PORT" "Agent Server" 60 &
WAIT_PID1=$!
wait_for_port "$AUTOMATION_PORT" "Automation Server" 60 &
WAIT_PID2=$!
wait "$WAIT_PID1" "$WAIT_PID2"
# ── 4. Start static server (frontend + proxy) ────────────────────────────────
log "Starting frontend + proxy on port $PORT..."
node /opt/agent-canvas/static-server.mjs \
--port "$PORT" \
--host 0.0.0.0 \
--dir /opt/agent-canvas/frontend \
--route "/api/automation=http://127.0.0.1:${AUTOMATION_PORT}" \
--route "/api=http://127.0.0.1:${AGENT_SERVER_PORT}" \
--route "/server_info=http://127.0.0.1:${AGENT_SERVER_PORT}" \
--route "/sockets=http://127.0.0.1:${AGENT_SERVER_PORT}" \
--route "/alive=http://127.0.0.1:${AGENT_SERVER_PORT}" \
--route "/health=http://127.0.0.1:${AGENT_SERVER_PORT}" \
--route "/ready=http://127.0.0.1:${AGENT_SERVER_PORT}" \
--route "/docs=http://127.0.0.1:${AGENT_SERVER_PORT}" \
--route "/redoc=http://127.0.0.1:${AGENT_SERVER_PORT}" \
--route "/openapi.json=http://127.0.0.1:${AGENT_SERVER_PORT}" &
PIDS+=($!)
log "All services started. Unified entry point: http://0.0.0.0:${PORT}/"
# Wait for any child to exit. If one dies, the trap will clean up the rest.
wait -n "${PIDS[@]}" 2>/dev/null
EXIT_CODE=$?
log_error "A service exited with code $EXIT_CODE"
exit "$EXIT_CODE"
+2 -1
View File
@@ -94,7 +94,8 @@
"typecheck:staged": "react-router typegen && npx tsc --noEmit --skipLibCheck",
"check-translation-completeness": "node scripts/check-translation-completeness.cjs",
"build:app": "npm run make-i18n && react-router build",
"build:lib": "npm run make-i18n && react-router typegen && cross-env BUILD_LIB=true VITE_APP_ENV=production vite build && tsc -p tsconfig.lib.json"
"build:lib": "npm run make-i18n && react-router typegen && cross-env BUILD_LIB=true VITE_APP_ENV=production vite build && tsc -p tsconfig.lib.json",
"build:docker": "node scripts/docker-build.mjs"
},
"lint-staged": {
"src/**/*.{ts,tsx,js}": [
+43 -57
View File
@@ -11,9 +11,9 @@
* - openhands-agent-server
*
* This script checks the RELEASED PyPI version of openhands-automation (as specified
* by DEFAULT_AUTOMATION_VERSION in dev-with-automation.mjs), not the main branch.
* DEFAULT_AUTOMATION_SDK_VERSION records the SDK dependency version for that
* released automation package and may intentionally lag DEFAULT_AGENT_SERVER_VERSION.
* by versions.automation in config/defaults.json), not the main branch.
* versions.automationSdk records the SDK dependency version for that
* released automation package and may intentionally lag versions.agentServer.
*
* This script is run in CI to catch version drift between projects.
*
@@ -23,9 +23,9 @@
* node scripts/check-sdk-version-sync.mjs --check-pypi
*
* Environment variables:
* EXPECTED_SDK_VERSION - Override the expected version (instead of reading from dev-with-automation.mjs)
* EXPECTED_SDK_VERSION - Override the expected version (instead of reading from config/defaults.json)
* AUTOMATION_PACKAGE_NAME - Override the automation package name (default: openhands-automation)
* AUTOMATION_PACKAGE_VERSION - Override the automation package version (instead of reading from dev-with-automation.mjs)
* AUTOMATION_PACKAGE_VERSION - Override the automation package version (instead of reading from config/defaults.json)
*
* Options:
* --check-pypi Also check the latest SDK version on PyPI
@@ -56,10 +56,9 @@ SDK Version Sync Check
Verifies that the released openhands-automation package on PyPI uses the
SDK version expected for that automation release.
The automation version is read from DEFAULT_AUTOMATION_VERSION in
dev-with-automation.mjs (currently used for local development). The expected
SDK dependency version is read from DEFAULT_AUTOMATION_SDK_VERSION when present,
falling back to DEFAULT_AGENT_SERVER_VERSION for older configs.
The automation version is read from config/defaults.json (versions.automation).
The expected SDK dependency version is read from versions.automationSdk,
falling back to versions.agentServer for older configs.
Usage:
node scripts/check-sdk-version-sync.mjs [options]
@@ -69,9 +68,9 @@ Options:
--help, -h Show this help
Environment variables:
EXPECTED_SDK_VERSION Override the expected SDK version (instead of reading from dev-with-automation.mjs)
EXPECTED_SDK_VERSION Override the expected SDK version (instead of reading from config/defaults.json)
AUTOMATION_PACKAGE_NAME Override the automation package name (default: openhands-automation)
AUTOMATION_PACKAGE_VERSION Override the automation package version (instead of reading from dev-with-automation.mjs)
AUTOMATION_PACKAGE_VERSION Override the automation package version (instead of reading from config/defaults.json)
Triggering from other repos:
The automation repo or SDK repo can trigger this check via GitHub repository_dispatch:
@@ -143,27 +142,31 @@ function sleep(ms) {
return new Promise((resolve) => setTimeout(resolve, ms));
}
/**
* Read the default agent-server SDK version from dev-safe.mjs.
*/
function getDefaultAgentServerVersion() {
const devSafePath = join(projectRoot, "scripts", "dev-safe.mjs");
const content = readFileSync(devSafePath, "utf8");
const match = content.match(
/const DEFAULT_AGENT_SERVER_VERSION = "([^"]+)"/,
// ── Centralized config ──────────────────────────────────────────────────────
let SHARED_DEFAULTS;
try {
SHARED_DEFAULTS = JSON.parse(
readFileSync(join(projectRoot, "config", "defaults.json"), "utf-8"),
);
if (!match) {
throw new Error(
"Could not find DEFAULT_AGENT_SERVER_VERSION in dev-safe.mjs",
);
if (!SHARED_DEFAULTS.versions?.agentServer || !SHARED_DEFAULTS.versions?.automationSdk) {
throw new Error("missing required fields: versions.agentServer, versions.automationSdk");
}
return { version: match[1], source: "dev-safe.mjs" };
} catch (err) {
console.error(`${colors.red}Failed to load config/defaults.json: ${err.message}${colors.reset}`);
console.error("Ensure the file exists and contains valid JSON with required fields.");
process.exit(1);
}
/**
* Read the expected automation SDK dependency version from environment,
* dev-with-automation.mjs, or dev-safe.mjs.
* Read the default agent-server SDK version from config/defaults.json.
*/
function getDefaultAgentServerVersion() {
return { version: SHARED_DEFAULTS.versions.agentServer, source: "config/defaults.json" };
}
/**
* Read the expected automation SDK dependency version from environment
* or config/defaults.json.
*/
function getExpectedVersion() {
// Allow override via environment variable (useful for CI triggers).
@@ -172,19 +175,10 @@ function getExpectedVersion() {
return { version: envVersion.trim(), source: "EXPECTED_SDK_VERSION env var" };
}
const devAutomationPath = join(projectRoot, "scripts", "dev-with-automation.mjs");
const content = readFileSync(devAutomationPath, "utf8");
const match = content.match(
/const DEFAULT_AUTOMATION_SDK_VERSION = "([^"]+)"/,
);
if (match) {
return {
version: match[1],
source: "DEFAULT_AUTOMATION_SDK_VERSION in dev-with-automation.mjs",
};
}
return getDefaultAgentServerVersion();
return {
version: SHARED_DEFAULTS.versions.automationSdk,
source: "config/defaults.json (versions.automationSdk)",
};
}
/**
@@ -205,7 +199,7 @@ async function fetchPyPIVersion(packageName) {
}
/**
* Read the automation version from env var or dev-with-automation.mjs
* Read the automation version from env var or config/defaults.json
*/
function getAutomationVersion() {
// Allow override via environment variable
@@ -214,18 +208,10 @@ function getAutomationVersion() {
return { version: envVersion.trim(), source: "AUTOMATION_PACKAGE_VERSION env var" };
}
const devAutomationPath = join(projectRoot, "scripts", "dev-with-automation.mjs");
const content = readFileSync(devAutomationPath, "utf8");
const match = content.match(
/const DEFAULT_AUTOMATION_VERSION = "([^"]+)"/,
);
if (!match) {
throw new Error(
"Could not find DEFAULT_AUTOMATION_VERSION in dev-with-automation.mjs",
);
}
return { version: match[1], source: "dev-with-automation.mjs" };
return {
version: SHARED_DEFAULTS.versions.automation,
source: "config/defaults.json (versions.automation)",
};
}
/**
@@ -324,7 +310,7 @@ async function main() {
console.log("");
try {
// Get expected version from env var or dev-safe.mjs
// Get expected version from env var or config/defaults.json
const { version: expectedVersion, source: versionSource } = getExpectedVersion();
console.log(
`Expected automation SDK version: ${colors.green}${expectedVersion}${colors.reset} (from ${versionSource})`,
@@ -337,7 +323,7 @@ async function main() {
);
}
// Get automation version from env var or dev-with-automation.mjs
// Get automation version from env var or config/defaults.json
const { version: automationVersion, source: automationSource } = getAutomationVersion();
console.log(
`Automation package: ${colors.cyan}${AUTOMATION_PACKAGE_NAME}==${automationVersion}${colors.reset} (from ${automationSource})`,
@@ -427,13 +413,13 @@ async function main() {
console.log("");
console.log("To fix, update one of the following:");
console.log(
` 1. Update DEFAULT_AUTOMATION_SDK_VERSION in scripts/dev-with-automation.mjs to match the automation release`,
` 1. Update versions.automationSdk in config/defaults.json to match the automation release`,
);
console.log(
` 2. Release a new version of ${AUTOMATION_PACKAGE_NAME} with SDK dependencies pinned to ${expectedVersion}`,
);
console.log(
` 3. Update DEFAULT_AUTOMATION_VERSION in scripts/dev-with-automation.mjs to a newer release`,
` 3. Update versions.automation in config/defaults.json to a newer release`,
);
console.log("");
process.exit(1);
+10 -8
View File
@@ -22,10 +22,16 @@ import {
signalProcessTree,
} from "./dev-process-utils.mjs";
const DEFAULT_BACKEND_PORT = 18000;
// ── Centralized config (single source of truth for versions, ports, etc.) ───
const __dev_safe_dirname = path.dirname(fileURLToPath(import.meta.url));
const SHARED_DEFAULTS = JSON.parse(
readFileSync(path.join(__dev_safe_dirname, "..", "config", "defaults.json"), "utf-8"),
);
const DEFAULT_BACKEND_PORT = SHARED_DEFAULTS.ports.agentServer;
const DEFAULT_VITE_PORT = 3001;
const DEFAULT_WAIT_TIMEOUT_MS = 30_000;
const DEFAULT_AGENT_SERVER_PACKAGE = "openhands-agent-server";
const DEFAULT_AGENT_SERVER_PACKAGE = SHARED_DEFAULTS.packages.agentServer;
const AGENT_SERVER_GIT_REPO = "https://github.com/OpenHands/software-agent-sdk";
const LOCAL_AGENT_SERVER_SUBDIRS = [
"openhands-agent-server",
@@ -33,12 +39,8 @@ const LOCAL_AGENT_SERVER_SUBDIRS = [
"openhands-tools",
"openhands-workspace",
];
// Default secret key for local development (DO NOT use in production)
// This is kept static because it's used for encrypting/decrypting persisted settings
const DEFAULT_SECRET_KEY = "openhands-dev-secret-key-change-in-prod";
// Default agent-server version (released PyPI version)
// Set OH_AGENT_SERVER_GIT_REF to use a git branch/SHA instead
const DEFAULT_AGENT_SERVER_VERSION = "1.22.1";
const DEFAULT_SECRET_KEY = SHARED_DEFAULTS.defaults.secretKey;
const DEFAULT_AGENT_SERVER_VERSION = SHARED_DEFAULTS.versions.agentServer;
const FRONTEND_REQUIRED_BINS = ["cross-env", "react-router"];
/**
+12 -9
View File
@@ -39,7 +39,7 @@
*/
import { spawn, spawnSync } from "node:child_process";
import { mkdirSync, existsSync } from "node:fs";
import { mkdirSync, existsSync, readFileSync } from "node:fs";
import { join, resolve, dirname } from "node:path";
import { fileURLToPath, pathToFileURL } from "node:url";
import { homedir } from "node:os";
@@ -67,16 +67,19 @@ import {
const __dirname = dirname(fileURLToPath(import.meta.url));
const projectRoot = resolve(__dirname, "..");
// ── Centralized config (single source of truth for versions, ports, etc.) ───
const SHARED_DEFAULTS = JSON.parse(
readFileSync(join(projectRoot, "config", "defaults.json"), "utf-8"),
);
const DEFAULT_AUTOMATION_REPO = "https://github.com/OpenHands/automation";
const DEFAULT_AUTOMATION_PACKAGE = "openhands-automation";
// Default automation version (released PyPI version)
// Set OH_AUTOMATION_GIT_REF to use a git branch/SHA instead
const DEFAULT_AUTOMATION_VERSION = "1.0.0a3";
const DEFAULT_AUTOMATION_PACKAGE = SHARED_DEFAULTS.packages.automation;
const DEFAULT_AUTOMATION_VERSION = SHARED_DEFAULTS.versions.automation;
// SDK version used by DEFAULT_AUTOMATION_VERSION. This can intentionally lag
// DEFAULT_AGENT_SERVER_VERSION while automation releases catch up.
const DEFAULT_AUTOMATION_SDK_VERSION = "1.22.1";
const DEFAULT_BACKEND_PORT = 18000;
const DEFAULT_AUTOMATION_PORT = 18001;
// the agent-server version while automation releases catch up.
const DEFAULT_AUTOMATION_SDK_VERSION = SHARED_DEFAULTS.versions.automationSdk;
const DEFAULT_BACKEND_PORT = SHARED_DEFAULTS.ports.agentServer;
const DEFAULT_AUTOMATION_PORT = SHARED_DEFAULTS.ports.automation;
// Where the auto-generated default automation API key is persisted. Static
// frontend builds bake VITE_AUTOMATION_API_KEY at build time, so the default
// must remain stable across restarts and --skip-build reuse.
+71
View File
@@ -0,0 +1,71 @@
#!/usr/bin/env node
/**
* Local Docker build helper.
*
* Reads version pins from config/defaults.json and invokes `docker build`
* with the correct --build-arg values so developers never need to remember
* (or hardcode) version strings.
*
* Usage:
* node scripts/docker-build.mjs # defaults
* node scripts/docker-build.mjs --tag my-tag # custom tag
* node scripts/docker-build.mjs -- --no-cache # extra docker args
*/
import { readFileSync } from "node:fs";
import { execFileSync } from "node:child_process";
import { fileURLToPath } from "node:url";
import { dirname, join } from "node:path";
const __dirname = dirname(fileURLToPath(import.meta.url));
const projectRoot = join(__dirname, "..");
const config = JSON.parse(
readFileSync(join(projectRoot, "config", "defaults.json"), "utf-8"),
);
const agentServerImage = `${config.images.agentServer}:${config.versions.agentServer}-python`;
const automationVersion = config.versions.automation;
// Parse CLI: --tag <name> and everything after -- is passed to docker build
let tag = "agent-canvas:local";
const extraArgs = [];
const args = process.argv.slice(2);
for (let i = 0; i < args.length; i++) {
if (args[i] === "--tag" && i + 1 < args.length) {
tag = args[++i];
} else if (args[i] === "--") {
extraArgs.push(...args.slice(i + 1));
break;
} else {
extraArgs.push(args[i]);
}
}
const cmd = [
"docker",
"build",
"-f",
"docker/Dockerfile",
"--build-arg",
`AGENT_SERVER_IMAGE=${agentServerImage}`,
"--build-arg",
`AUTOMATION_VERSION=${automationVersion}`,
"-t",
tag,
...extraArgs,
".",
];
console.log(`Agent Server image : ${agentServerImage}`);
console.log(`Automation version : ${automationVersion}`);
console.log(`Tag : ${tag}`);
console.log(`\n$ ${cmd.join(" ")}\n`);
try {
execFileSync(cmd[0], cmd.slice(1), {
cwd: projectRoot,
stdio: "inherit",
});
} catch (err) {
process.exit(err.status || 1);
}
+4 -8
View File
@@ -78,10 +78,8 @@ const conversations: MockConversation[] = [
{
id: PAGINATION_LOCAL_CONVERSATION_ID,
title: "Local pagination fixture",
created_at: new Date(PAGINATION_BASE_TIME).toISOString(),
updated_at: new Date(
PAGINATION_BASE_TIME + PAGINATION_EVENT_COUNT * 60_000,
).toISOString(),
created_at: new Date(now - 6 * 24 * 60 * 60 * 1000).toISOString(),
updated_at: new Date(now - 6 * 24 * 60 * 60 * 1000).toISOString(),
execution_status: "idle",
workspace: { working_dir: "/workspace/project" },
},
@@ -160,10 +158,8 @@ async function maybeReturnPaginationEvents(
}
function createCloudPaginationConversation(): AppConversation {
const createdAt = new Date(PAGINATION_BASE_TIME).toISOString();
const updatedAt = new Date(
PAGINATION_BASE_TIME + PAGINATION_EVENT_COUNT * 60_000,
).toISOString();
const createdAt = new Date(now - 6 * 24 * 60 * 60 * 1000).toISOString();
const updatedAt = createdAt;
return {
id: PAGINATION_CLOUD_CONVERSATION_ID,