Remove the local relay dashboard#1146
Merged
Merged
Conversation
The local web dashboard (relay-dashboard-server / @agent-relay/dashboard-server) is gone. `agent-relay up` is now a plain broker launcher — attached by default, `--background` to detach — and the dashboard-only flags (`--no-dashboard`, `--port`, `--foreground`) are removed. - broker-lifecycle: drop dashboard spawning, static-asset discovery/refresh, the GitHub dashboard-ui download, the dead reuse-broker branch, and dashboard-port logic. Add `AGENT_RELAY_BROKER_PORT` (resolveBrokerBasePort) for the broker base port; orphan detection now keys on `up` minus `--background` instead of the removed `--foreground`. - core / core-maintenance: remove the dashboard flags, binary resolver, and uninstall handling for dashboard assets/binary/npm package. - config: drop the unused localDashboardUrl / LOCAL_DASHBOARD_URL. - telemetry: drop the `dashboard` / `human_dashboard` enum members. - fleet + e2e harness: AGENT_RELAY_DASHBOARD_PORT -> AGENT_RELAY_BROKER_PORT. - scripts: delete the dashboard launcher and the stale sibling-repo test script; fix removed-flag usages in CI/e2e/demo scripts. - docs: drop dashboard wording and removed flags from the current CLI docs. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
Warning You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again! |
Contributor
|
Caution Review failedPull request was closed or merged during review No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (3)
🚧 Files skipped from review as they are similar to previous changes (1)
📝 WalkthroughWalkthroughRemoves the dashboard component from the CLI broker lifecycle: the ChangesDashboard removal from CLI broker lifecycle
Sequence Diagram(s)sequenceDiagram
participant runUpCommand
participant resolveBrokerBasePort
participant startBrokerWithPortFallback
participant classifyBrokerStartStage
runUpCommand->>resolveBrokerBasePort: reads AGENT_RELAY_BROKER_PORT env
resolveBrokerBasePort-->>runUpCommand: basePort (default 3888)
alt background spawn
runUpCommand->>runUpCommand: filter argv remove --background
runUpCommand->>runUpCommand: spawnProcess detached
else foreground startup
runUpCommand->>startBrokerWithPortFallback: basePort
startBrokerWithPortFallback-->>runUpCommand: apiPort = basePort + 1 (free port scan)
runUpCommand->>runUpCommand: log relay API URL
runUpCommand->>runUpCommand: setup fleet sidecar best-effort
end
runUpCommand->>runUpCommand: error path
runUpCommand->>classifyBrokerStartStage: classify failure (err, message)
classifyBrokerStartStage-->>runUpCommand: stage label
runUpCommand->>runUpCommand: telemetry broker_start_failed
Estimated code review effort🎯 4 (Complex) | ⏱️ ~75 minutes Possibly related PRs
Suggested labels
Suggested reviewers
Poem
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Contributor
|
Preview deployed!
This preview will be cleaned up when the PR is merged or closed. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: f8d8cb6688
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| echo "=== Smoke: standalone up ===" | ||
| UP_LOG="$TMP_ROOT/up.log" | ||
| run_cli local up --no-dashboard >"$UP_LOG" 2>&1 & | ||
| run_cli local up >"$UP_LOG" 2>&1 & |
There was a problem hiding this comment.
In this smoke path, local up now stays attached by default, but the script still backgrounds it and later does local down followed by wait "$UP_PID". down stops the broker process recorded in connection.json, not this attached CLI wrapper that is stuck in holdOpen(), so the smoke job hangs after a successful start. Use local up --background here to preserve the old detached --no-dashboard behavior.
Useful? React with 👍 / 👎.
| "build:telemetry": "npm --prefix packages/telemetry run build", | ||
| "dev:watch": "cd packages/cli && npx tsc -w", | ||
| "watch:start": "npm run build && concurrently -k \"npm run dev:watch\" \"node --watch packages/cli/dist/cli/index.js start dashboard.js claude\"", | ||
| "watch:start": "npm run build && concurrently -k \"npm run dev:watch\" \"node --watch packages/cli/dist/cli/index.js up\"", |
There was a problem hiding this comment.
When npm run watch:start runs this command, Commander receives up as a top-level command, but createProgram() registers broker lifecycle commands only under local (local up, local down, etc.). The watcher therefore exits with an unknown-command error instead of starting the broker; this should call node --watch packages/cli/dist/cli/index.js local up.
Useful? React with 👍 / 👎.
|
|
||
| // Remove standalone binaries from ~/.local/bin | ||
| for (const binaryName of ['agent-relay', 'relay-dashboard-server', 'relay-acp']) { | ||
| for (const binaryName of ['agent-relay', 'relay-acp']) { |
There was a problem hiding this comment.
For users installed by the current install.sh, the installer can still place relay-dashboard-server in the install bin directory or fall back to the global @agent-relay/dashboard-server package, but this uninstall path now removes only agent-relay and relay-acp. In those installs, agent-relay uninstall leaves the dashboard binary/package/assets behind; keep removing the dashboard artifacts until the installer and old-install migration no longer produce them.
Useful? React with 👍 / 👎.
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 3
🧹 Nitpick comments (1)
scripts/watch-cli-tools.sh (1)
21-24: ⚡ Quick winWarn on ignored flags instead of silently swallowing them.
Ignoring unknown args without feedback can hide typos and mis-invocations during local dev.
Suggested patch
*) - # Other flags (e.g. --tool) are accepted for compatibility but ignored. - : + # Other flags (e.g. --tool/--port) are accepted for compatibility but ignored. + echo "[watch-cli-tools] Ignoring flag: $1" >&2 ;;🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@scripts/watch-cli-tools.sh` around lines 21 - 24, The default case in the watch-cli-tools.sh script currently uses a colon (no-op) to silently ignore unknown flags. Replace the `:` operator in the `*)` case with an echo or printf statement that outputs a warning message to stderr (using `>&2` redirection) to alert users when they pass unrecognized flags, helping catch typos and mis-invocations during local development.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@packages/cli/src/cli/lib/broker-lifecycle.ts`:
- Around line 220-223: The resolveBrokerBasePort function validates that
AGENT_RELAY_BROKER_PORT is a positive integer but lacks an upper bound check,
allowing invalid values like 65535 which would cause port allocation failures
downstream. Modify the return statement in resolveBrokerBasePort to also
validate that the parsed port does not exceed 65534 (to account for port
increments that occur later in port allocation). If the port exceeds this
maximum, return DEFAULT_BROKER_BASE_PORT or throw a configuration error with a
clear message indicating the valid port range.
In `@packages/cli/src/cli/lib/core-maintenance.ts`:
- Line 287: The uninstall process is removing cleanup logic for legacy
dashboard-era artifacts including the agent-relay and relay-acp binaries, which
will leave orphaned files for upgraded users. Retain the cleanup code for these
legacy binaries (referenced in the loop iterating over ['agent-relay',
'relay-acp'] and the related removal logic around lines 322-331) as best-effort
backward-compatible cleanup in the uninstall function, ensuring users don't end
up with orphaned global binaries after upgrading and uninstalling.
In `@scripts/e2e-test.sh`:
- Around line 154-157: The preflight cleanup block with the lsof command is only
targeting the base $BROKER_PORT, but the broker API actually binds to
$BROKER_PORT + 1. Modify the lsof command to kill processes on both ports to
ensure proper cleanup and isolation. Update the logic to kill processes on both
the base port ($BROKER_PORT) and the API port ($BROKER_PORT + 1) in the same
cleanup block.
---
Nitpick comments:
In `@scripts/watch-cli-tools.sh`:
- Around line 21-24: The default case in the watch-cli-tools.sh script currently
uses a colon (no-op) to silently ignore unknown flags. Replace the `:` operator
in the `*)` case with an echo or printf statement that outputs a warning message
to stderr (using `>&2` redirection) to alert users when they pass unrecognized
flags, helping catch typos and mis-invocations during local development.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro Plus
Run ID: 66bf73ad-514f-4d36-8b37-ee066f137fd4
📒 Files selected for processing (28)
package.jsonpackages/cli/src/cli/bootstrap.test.tspackages/cli/src/cli/bootstrap.tspackages/cli/src/cli/commands/core.test.tspackages/cli/src/cli/commands/core.tspackages/cli/src/cli/commands/fleet.tspackages/cli/src/cli/commands/local-agent.tspackages/cli/src/cli/lib/broker-lifecycle.test.tspackages/cli/src/cli/lib/broker-lifecycle.tspackages/cli/src/cli/lib/core-maintenance.tspackages/cli/src/cli/telemetry/events.tspackages/config/src/cloud-config.tspackages/config/src/schemas.tspackages/harness-driver/src/spawn-config.tspackages/policy/src/agent-policy.tsscripts/ci-standalone-smoke.shscripts/demos/server-capacity.shscripts/demos/sprint-planning.shscripts/e2e-test.shscripts/run-dashboard.jsscripts/test-interactive-terminal.shscripts/test-spawn-refactor.shscripts/watch-cli-tools.shtests/e2e/fleet/fleet-e2e.test.tstests/e2e/fleet/harness.tsweb/content/docs/cli-broker-lifecycle.mdxweb/content/docs/cli-overview.mdxweb/content/docs/reference-cli.mdx
💤 Files with no reviewable changes (3)
- scripts/run-dashboard.js
- scripts/test-spawn-refactor.sh
- packages/config/src/schemas.ts
Comment on lines
+220
to
+223
| export function resolveBrokerBasePort(deps: Pick<CoreDependencies, 'env'>): number { | ||
| const raw = Number.parseInt(deps.env.AGENT_RELAY_BROKER_PORT ?? '', 10); | ||
| return Number.isFinite(raw) && raw > 0 ? raw : DEFAULT_BROKER_BASE_PORT; | ||
| } |
Contributor
There was a problem hiding this comment.
Validate upper bound for AGENT_RELAY_BROKER_PORT to prevent impossible API port allocation.
Line 221 currently accepts any positive integer. If a user sets AGENT_RELAY_BROKER_PORT=65535, Line 234 computes 65536, and startup fails every time. Clamp to a valid base-port range (1..65534) or fail fast with a clear config error.
Suggested fix
export function resolveBrokerBasePort(deps: Pick<CoreDependencies, 'env'>): number {
const raw = Number.parseInt(deps.env.AGENT_RELAY_BROKER_PORT ?? '', 10);
- return Number.isFinite(raw) && raw > 0 ? raw : DEFAULT_BROKER_BASE_PORT;
+ const maxBasePort = MAX_PORT - 1; // API binds to basePort + 1
+ return Number.isFinite(raw) && raw > 0 && raw <= maxBasePort ? raw : DEFAULT_BROKER_BASE_PORT;
}🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@packages/cli/src/cli/lib/broker-lifecycle.ts` around lines 220 - 223, The
resolveBrokerBasePort function validates that AGENT_RELAY_BROKER_PORT is a
positive integer but lacks an upper bound check, allowing invalid values like
65535 which would cause port allocation failures downstream. Modify the return
statement in resolveBrokerBasePort to also validate that the parsed port does
not exceed 65534 (to account for port increments that occur later in port
allocation). If the port exceeds this maximum, return DEFAULT_BROKER_BASE_PORT
or throw a configuration error with a clear message indicating the valid port
range.
|
|
||
| // Remove standalone binaries from ~/.local/bin | ||
| for (const binaryName of ['agent-relay', 'relay-dashboard-server', 'relay-acp']) { | ||
| for (const binaryName of ['agent-relay', 'relay-acp']) { |
Contributor
There was a problem hiding this comment.
Keep best-effort legacy dashboard cleanup in uninstall.
This removes cleanup for legacy dashboard-era artifacts, so upgraded users can be left with orphaned global binaries/packages after uninstall. Keep these removals as backward-compatible best-effort cleanup.
Suggested patch
- for (const binaryName of ['agent-relay', 'relay-acp']) {
+ for (const binaryName of ['agent-relay', 'relay-acp', 'relay-dashboard-server']) {
@@
- for (const pkg of ['agent-relay']) {
+ for (const pkg of ['agent-relay', '`@agent-relay/dashboard-server`']) {
@@
- deps.log('[dry-run] Would run: npm uninstall -g agent-relay');
+ deps.log('[dry-run] Would run: npm uninstall -g agent-relay');
+ deps.log('[dry-run] Would run: npm uninstall -g `@agent-relay/dashboard-server`');Also applies to: 322-331
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@packages/cli/src/cli/lib/core-maintenance.ts` at line 287, The uninstall
process is removing cleanup logic for legacy dashboard-era artifacts including
the agent-relay and relay-acp binaries, which will leave orphaned files for
upgraded users. Retain the cleanup code for these legacy binaries (referenced in
the loop iterating over ['agent-relay', 'relay-acp'] and the related removal
logic around lines 322-331) as best-effort backward-compatible cleanup in the
uninstall function, ensuring users don't end up with orphaned global binaries
after upgrading and uninstalling.
Comment on lines
+154
to
157
| # Kill any process using our target port (ensures the broker can bind) | ||
| if command -v lsof &> /dev/null; then | ||
| lsof -ti:$DASHBOARD_PORT | xargs kill -9 2>/dev/null || true | ||
| lsof -ti:$BROKER_PORT | xargs kill -9 2>/dev/null || true | ||
| fi |
Contributor
There was a problem hiding this comment.
Preflight cleanup is targeting the base port, not the broker API port.
With AGENT_RELAY_BROKER_PORT, the API binds from base+1; killing only $BROKER_PORT can leave stale API listeners and make E2E runs less isolated.
Suggested patch
-# Kill any process using our target port (ensures the broker can bind)
+# Kill any process using the expected broker API port (base+1)
if command -v lsof &> /dev/null; then
- lsof -ti:$BROKER_PORT | xargs kill -9 2>/dev/null || true
+ API_PORT=$((BROKER_PORT + 1))
+ lsof -ti:$API_PORT | xargs kill -9 2>/dev/null || true
fi📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| # Kill any process using our target port (ensures the broker can bind) | |
| if command -v lsof &> /dev/null; then | |
| lsof -ti:$DASHBOARD_PORT | xargs kill -9 2>/dev/null || true | |
| lsof -ti:$BROKER_PORT | xargs kill -9 2>/dev/null || true | |
| fi | |
| # Kill any process using the expected broker API port (base+1) | |
| if command -v lsof &> /dev/null; then | |
| API_PORT=$((BROKER_PORT + 1)) | |
| lsof -ti:$API_PORT | xargs kill -9 2>/dev/null || true | |
| fi |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@scripts/e2e-test.sh` around lines 154 - 157, The preflight cleanup block with
the lsof command is only targeting the base $BROKER_PORT, but the broker API
actually binds to $BROKER_PORT + 1. Modify the lsof command to kill processes on
both ports to ensure proper cleanup and isolation. Update the logic to kill
processes on both the base port ($BROKER_PORT) and the API port ($BROKER_PORT +
1) in the same cleanup block.
Two CI failures on the dashboard-removal branch: - package-validation "Build & Validate" called `local up --port`, a flag removed with the dashboard. Route the test's port through AGENT_RELAY_BROKER_PORT and start a plain `local up` instead. - Standalone macOS Smoke crashed: `defineDefaultLocalNode is not a function`. bun's `--compile` resolves @agent-relay/fleet's value exports to its type-only .d.ts, so the implicit fleet node helpers are undefined in the standalone binary. This is pre-existing; it was hidden because `up --no-dashboard` previously ran the fleet sidecar in a detached child. Attached `up` surfaced it. The implicit local fleet node is best-effort, so wrap its startup and warn-and-continue instead of aborting the broker. Also drop the now-dead "Download dashboard binary" step from e2e-tests.yml. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In @.github/workflows/package-validation.yml:
- Around line 196-203: The health check hardcodes the API_PORT as BROKER_PORT +
1, but the broker startup can select the next available port if that one is
occupied, causing the curl request to fail against the wrong endpoint. Instead
of assuming a fixed offset, capture the actual API port that the broker selects
by parsing the broker's startup output or having it write the port to a file,
then use that captured value in the curl health check loop. This ensures the
health check queries the correct port regardless of what the broker actually
bound to.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro Plus
Run ID: c4a689ff-4ed5-4b5f-87b8-71b61c1d699a
📒 Files selected for processing (3)
.github/workflows/e2e-tests.yml.github/workflows/package-validation.ymlpackages/cli/src/cli/lib/broker-lifecycle.ts
💤 Files with no reviewable changes (1)
- .github/workflows/e2e-tests.yml
🚧 Files skipped from review as they are similar to previous changes (1)
- packages/cli/src/cli/lib/broker-lifecycle.ts
Comment on lines
+196
to
203
| API_PORT=$((BROKER_PORT + 1)) | ||
| # Start the broker in background | ||
| AGENT_RELAY_BROKER_PORT="$BROKER_PORT" node packages/cli/dist/cli/index.js local up & | ||
| DAEMON_PID=$! | ||
|
|
||
| # Wait for health endpoint (broker API is dashboard port + 1) | ||
| # Wait for health endpoint (broker API is the base port + 1) | ||
| for i in {1..20}; do | ||
| if curl -sf "http://127.0.0.1:${API_PORT}/health" > /dev/null; then |
Contributor
There was a problem hiding this comment.
Health check hardcodes base+1 even though broker port selection can fall back
Line 196/Line 203 assume the API will always bind to BROKER_PORT + 1, but lifecycle startup may move to the next free port when that slot is occupied. This can fail CI even when the broker started successfully.
Suggested fix
BROKER_PORT=3899
- API_PORT=$((BROKER_PORT + 1))
+ START_API_PORT=$((BROKER_PORT + 1))
+ API_PORT=""
# Start the broker in background
AGENT_RELAY_BROKER_PORT="$BROKER_PORT" node packages/cli/dist/cli/index.js local up &
DAEMON_PID=$!
- # Wait for health endpoint (broker API is the base port + 1)
+ # Wait for health endpoint (API starts at base+1 and may fall back)
for i in {1..20}; do
- if curl -sf "http://127.0.0.1:${API_PORT}/health" > /dev/null; then
- break
- fi
+ for candidate in $(seq "$START_API_PORT" "$((START_API_PORT + 9))"); do
+ if curl -sf "http://127.0.0.1:${candidate}/health" > /dev/null; then
+ API_PORT="$candidate"
+ break 2
+ fi
+ done
sleep 1
done
# Verify broker is reachable
- if curl -sf "http://127.0.0.1:${API_PORT}/health" > /dev/null; then
+ if [ -n "$API_PORT" ] && curl -sf "http://127.0.0.1:${API_PORT}/health" > /dev/null; then🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In @.github/workflows/package-validation.yml around lines 196 - 203, The health
check hardcodes the API_PORT as BROKER_PORT + 1, but the broker startup can
select the next available port if that one is occupied, causing the curl request
to fail against the wrong endpoint. Instead of assuming a fixed offset, capture
the actual API port that the broker selects by parsing the broker's startup
output or having it write the port to a file, then use that captured value in
the curl health check loop. This ensures the health check queries the correct
port regardless of what the broker actually bound to.
# Conflicts: # packages/cli/src/cli/lib/broker-lifecycle.ts # web/content/docs/cli-overview.mdx
- install.sh: drop the dashboard-server binary download, dashboard UI download, `npm install -g @agent-relay/dashboard-server`, the AGENT_RELAY_NO_DASHBOARD env, the REPO_DASHBOARD repo, and the `up --dashboard` post-install help. The installer no longer installs the removed local dashboard. - Fix broken `--no-dashboard` / `up --dashboard` examples in using-agent-relay skill docs and the demos README; rename "dashboard port" -> "broker port" in the fleet e2e README; drop the stale run-dashboard.js line from .npmignore. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@install.sh`:
- Around line 687-693: The quick start documentation in the echo statements
shows agent-relay up being run without background mode, which will block the
terminal and prevent the subsequent status and down commands from executing.
Update the echo statements to either add the --background flag to the
agent-relay up command (changing it to agent-relay up --background), or add a
comment in the documentation explaining that users need to open a second
terminal for the status and down commands. This ensures the quick start flow is
actually executable as documented to users.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro Plus
Run ID: 912f40fd-55ce-4264-bc76-d44f7c96396d
📒 Files selected for processing (14)
.agents/skills/using-agent-relay/SKILL.md.claude/skills/using-agent-relay/SKILL.md.npmignoreinstall.shpackage.jsonpackages/cli/src/cli/bootstrap.test.tspackages/cli/src/cli/commands/fleet.tspackages/cli/src/cli/commands/local-agent.tspackages/utils/cli-registry.yamlscripts/demos/README.mdtests/e2e/fleet/README.mdweb/content/agent-relay/SKILL.mdweb/content/docs/cli-overview.mdxweb/content/docs/reference-cli.mdx
💤 Files with no reviewable changes (1)
- .npmignore
✅ Files skipped from review due to trivial changes (7)
- packages/cli/src/cli/bootstrap.test.ts
- web/content/docs/reference-cli.mdx
- packages/utils/cli-registry.yaml
- tests/e2e/fleet/README.md
- web/content/docs/cli-overview.mdx
- scripts/demos/README.md
- packages/cli/src/cli/commands/local-agent.ts
🚧 Files skipped from review as they are similar to previous changes (2)
- package.json
- packages/cli/src/cli/commands/fleet.ts
The local dashboard is gone, and the TS `ActionSource` union already dropped `human_dashboard`. Re-sync the Rust mirror: - Remove the `HumanDashboard` enum variant + its `as_str` arm so the broker's taxonomy matches TS (`human_cli | agent | protocol`). - Re-point the only emitter — the `/api/spawn` HTTP handler — to `ActionSource::HumanCli`. That endpoint's human caller is now the CLI (the dashboard GUI that previously posted to it is gone). - Update the telemetry unit tests. Shared broker infrastructure that is merely *named* "dashboard" (the `/ws` SDK event transport, `display_target_for_dashboard`, `sender_is_dashboard_label`, the operator "Dashboard" label) is left intact — it powers the SDK/CLI event stream and the relaycast/Observer message path, not the removed local UI. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…nable `agent-relay up` runs attached by default, which blocks the terminal and stops the documented `status`/`down` steps from running. Use `--background` so the quick-start sequence is executable as printed. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
willwashburn
added a commit
that referenced
this pull request
Jun 21, 2026
Add Removed / Breaking Changes / Migration Guidance entries to [Unreleased] for the local web dashboard removal landed in #1146: dropped `up` flags (--no-dashboard/--port/--foreground), AGENT_RELAY_DASHBOARD_PORT -> AGENT_RELAY_BROKER_PORT, installer no longer fetches the dashboard, and the `human_dashboard` ActionSource is retired. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Removes the local web dashboard (
relay-dashboard-server/@agent-relay/dashboard-server) and every reference/dependency on it across the CLI, supporting packages, helper scripts, and current CLI docs.agent-relay upis now a plain broker launcher:--backgroundto detach--no-dashboard,--port, and--foregroundare goneChanges
CLI / broker (
packages/cli)broker-lifecycle.ts: removed all dashboard machinery (spawning, static-asset discovery/refresh, the GitHubdashboard-ui.tar.gzdownload, dashboard port-fallback) and the dead "reuse broker for dashboard" branch. AddedresolveBrokerBasePort()backed byAGENT_RELAY_BROKER_PORT(default 3888; broker API binds base+1). Orphan detection now keys onupminus--backgroundinstead of the removed--foreground.core.ts: dropped the dashboard flags, the dashboard-binary resolver, and the unused dashboard-portfindBrokerApiPort.core-maintenance.ts: uninstall no longer touches dashboard assets, therelay-dashboard-serverbinary, or the@agent-relay/dashboard-servernpm package.telemetry/events.ts: dropped thedashboard/human_dashboardenum members.Other packages:
config(removed the deadlocalDashboardUrl/LOCAL_DASHBOARD_URL); comment fixes inharness-driverandpolicy.Scripts: deleted
run-dashboard.jsand the stale sibling-repotest-spawn-refactor.sh; fixed removed-flag usages inci-standalone-smoke.sh,e2e-test.sh(its--portnow routes throughAGENT_RELAY_BROKER_PORT, so the CI workflow keeps working), the demos, andwatch-cli-tools.sh.Tests: updated
core.test.ts,broker-lifecycle.test.ts, and the e2e fleet harness for the new flag/port model.Docs: removed the dashboard wording and the
--no-dashboard/--port/--foregroundrows from the current CLI docs (cli-broker-lifecycle.mdx,reference-cli.mdx,cli-overview.mdx). The7.1.1/snapshot (incl.relay-dashboard.mdx) is intentionally left as a historical version; generic uses of the word "dashboards" (operator consoles / API consumers) are left intact. The clouddashboardUrl/appUrl(hosted web app) are a different concept and untouched.Verification
npm run typecheck— passes (exit 0)