What hyperdxio/hyperdx shipped
Written by FoxPlug from the repository's own releases; not affiliated with Hyperdx. Paste your own repository in the box below and FoxPlug writes the same weekly update for your product. This page is an automatic summary of the releases, pull requests and commits in github.com/hyperdxio/hyperdx; Hyperdx did not write it and does not use or endorse FoxPlug. Every line links to the change it describes.
Get a weekly update like this for your product, free
Week of November 2, 2026
What shipped
- The project now self-instruments with the ClickStack browser and Node SDKs. Commit
- Trimming tool responses in mcp now shortens content instead of replacing it. Commit
- The README and docs now refer to HyperDX as ClickStack. Commit
- The api systeminformation dependency was updated to 5.33.19. Commit
- The e2e app webServer startup timeout was raised to 6 minutes. Commit
- Fork-PR e2e shards now run on ubuntu-24.04 in CI. Commit
- CODEOWNERS was updated to reflect the new team structure. Commit
Why it matters
This week brings a new self-instrumentation path using the ClickStack browser and Node SDKs alongside a fix that keeps trimmed tool responses readable. The docs and README now use the ClickStack name, and test, CI and dependency updates keep the pipeline steady.
Changelog entry
- Self-instrument with the ClickStack browser and Node SDKs. Commit
- Shorten instead of replacing when trimming tool responses in mcp. Commit
- Rename HyperDX to ClickStack in README and docs. Commit
- Update systeminformation to 5.33.19 in api. Commit
- Raise app webServer startup timeout to 6 minutes for e2e. Commit
- Run fork-PR e2e shards on ubuntu-24.04 in CI. Commit
- Update CODEOWNERS to reflect new team structure. Commit
This week: self-instrumentation with the ClickStack browser and Node SDKs, tool response trimming now shortens instead of replacing, and the README and docs use the ClickStack name. Plus a systeminformation update, a longer e2e startup timeout, and CODEOWNERS refresh.
This week in the repository: the project now self-instruments with the ClickStack browser and Node SDKs, and trimming of tool responses shortens content instead of replacing it. The README and docs now refer to HyperDX as ClickStack. Smaller changes include updating the api systeminformation dependency to 5.33.19, raising the e2e app webServer startup timeout to 6 minutes, running fork-PR e2e shards on ubuntu-24.04, and updating CODEOWNERS to reflect the new team structure.
Week of October 26, 2026
What shipped
- Alert tags are now available as template variables in Generic and incident.io webhook bodies, letting shared webhooks route alerts by team. Release
- Session replay now renders browser events from OpenTelemetry SDKs in addition to the legacy HyperDX browser SDK. Pull request #3440
- Session replay reads the OpenTelemetry session.id semantic convention to display sessions from OTel browser SDKs. Pull request #3438
- ClickHouse bundled image pinned to 26.8.36.3-alpine to include recent security fixes. Pull request #3497
- Multipath TCP disabled in the OTel collector to mitigate CVE-2026-98432. Pull request #3505
- Cached table columns now expire after 5 minutes instead of persisting for the process lifetime. Pull request #3481
- Alert preview chart now highlights the correct region above the threshold for above-type alerts. Pull request #3488
- String literals in key-value filter rewrites are now preserved verbatim to fix row lookups. Pull request #3484
- Sidebar navigation labels changed to sentence case throughout the UI. Pull request #3489
- CODEOWNERS added to automatically assign pull reviews to the clickstack-app-devs team. Pull request #3493
Why it matters
Alert routing improvements let teams organize incoming alerts by owner through webhooks. Session replay support for OpenTelemetry browser SDKs enables tracking of sessions from the standard SDK ecosystem. Security updates for ClickHouse and the OTel collector address recent CVEs.
Changelog entry
- Alert tags exposed to Generic and incident.io webhook bodies as {{tags}} template variable Release
- Browser events now render in session replay from OpenTelemetry log-based EventName SDKs Pull request #3440
- Session replay reads OpenTelemetry session.id semantic convention Pull request #3438
- ClickHouse bundled image pinned to 26.8.36.3-alpine Pull request #3497
- Multipath TCP disabled in OTel collector to mitigate CVE-2026-98432 Pull request #3505
- Cached table columns expire after 5 minutes instead of persisting Pull request #3481
- Alert preview chart highlights correct region above threshold Pull request #3488
- String literals in KV filter rewrites preserved verbatim Pull request #3484
- Fixed turbopack failure in local development via next.js upgrade Pull request #3487
- Sidebar navigation labels changed to sentence case Pull request #3489
- Smoke tests ClickHouse updated to 26.8 Pull request #3498
- Release workflow Bun bumped to 1.4.4 for CVE-2026-111497 fix Pull request #3471
2.43 shipped: alert tags in webhooks for team routing, OTel session replay support, ClickHouse security fixes, and multipath TCP disabled in the collector.
2.43 is out. Alert tags now appear in Generic and incident.io webhook bodies for flexible routing. Session replay reads OpenTelemetry session IDs and browser events from OTel SDKs. ClickHouse pinned to a patched LTS version. Multipath TCP disabled in the collector for CVE mitigation.
Week of October 19, 2026
What shipped
- PromQL tiles in dashboards are now returned by the external API and accepted on create and update, so saving a dashboard back through the API no longer drops them. Pull request #3379
- The external API at /api/v2/dashboards now supports PromQL tiles, returning them and accepting them on create and update with configType promql, a sourceId and a list of expressions. Release
- The CLI adds dashboard and saved-search authoring commands, including dashboards list with name filtering and table, JSON or CSV output and dashboards export by id or name. Release
- Chart alerts can now be edited in a side panel docked beside the chart editor, so the preview chart stays in view while the alert is tuned. Release
- User preferences replace the Use UTC time switch with a searchable time zone selector covering local browser time, UTC and every zone the browser knows. Pull request #3449
- Log sources gain an optional SeverityNumber expression, so the numeric OpenTelemetry severity can be used instead of reading only SeverityText. Pull request #3425
- Webhook bodies for grouped alerts can reference the breaching group's individual values through variable placeholders instead of only the joined group key string. Pull request #3384
- Table charts support click-to-adhoc-filter across builder, raw SQL and PromQL tables, adding the clicked column's labels to the adhoc filter. Pull request #3389
- Time chart series tooltips gain an action that adds all of a series' labels to the relevant adhoc dashboard filter. Pull request #3405
- Node images are bumped to the fixed 22.23.4 line for CVE-2026-111497, an out-of-bounds write in the vendored brotli decoder, for which a proof of concept is public. Pull request #3456
Why it matters
This week adds PromQL dashboard tiles to the external API and closes the gap that deleted them on save, alongside CLI commands for authoring dashboards and saved searches without the web UI. Security work bumped Node images to a fixed line for a public brotli decoder vulnerability, and several fixes make alerts, exceptions and saved searches behave more consistently.
Changelog entry
- PromQL dashboard tiles are returned by the external API and accepted on create and update. Pull request #3379
- External API dashboards endpoint supports PromQL tiles with configType promql, a sourceId and expressions. Release
- CLI adds dashboard list with name filtering and format options plus dashboard export by id or name. Release
- Chart alerts can be edited in a side panel docked beside the chart editor. Release
- User preferences replace the Use UTC time switch with a searchable time zone selector. Pull request #3449
- Log sources gain an optional SeverityNumber expression for the numeric OpenTelemetry severity. Pull request #3425
- Grouped alert webhook bodies can reference individual group values instead of only the joined group key. Pull request #3384
- Table charts support click-to-adhoc-filter for builder, raw SQL and PromQL tables. Pull request #3389
- Time chart series tooltips can add all of a series' labels to the adhoc dashboard filter. Pull request #3405
- OTel exceptions appear in the side panel and trace waterfall without extra source configuration. Pull request #3412
- Node images bumped to the fixed 22.23.4 line for CVE-2026-111497 in the vendored brotli decoder. Pull request #3456
- Saved searches are usable on phones below 900px, with wrapping query rows and scrollable results. Pull request #3376
This week: PromQL dashboard tiles now round-trip through the external API, the CLI can list, export and author dashboards and saved searches, chart alerts edit in a side panel, and user preferences get a time zone selector. Node images bumped to the fixed line for CVE-2026-111497
This week in Hyperdx: PromQL dashboard tiles are now returned by the external API and accepted on create and update, so they are no longer lost when a dashboard is saved back. The CLI gained dashboard and saved-search authoring commands, including list with name filtering and export by id or name. Chart alerts can be edited in a side panel beside the chart editor, user preferences now use a searchable time zone selector, log sources gained a SeverityNumber expression, table charts support click-to-adhoc-filter, and Node images were bumped to the fixed 22.23.4 line for CVE-2026-111497.
Week of October 12, 2026
What shipped
- Hyperdx app 2.41.0 adds a series mode to builder heatmaps, PromQL series-mode heatmaps, and a minimum auto granularity setting for PromQL sources. Release
- Hyperdx API 2.41.0 stops falling back to the public session secret; if the secret is unset, the API generates a random one and reports it once at startup when authentication is enabled. Release
- Hyperdx CLI 0.7.0 adds a stacked line time-series display type. Release
- Prometheus histograms can now be displayed as heatmaps using a new histogram mode for pre-bucketed distribution data. Pull request #3332
- Heatmaps can now show a distribution calculated from PromQL values, bucketed client-side into linear or log-spaced buckets. Pull request #3326
- Adhoc dashboard filter conditions are now broadcast to builder tiles, SQL tiles through the filters macro, and PromQL expressions. Pull request #3366
- Filter queries no longer wait for all values: keys are shown in the UI as they arrive and values stream in afterwards. Pull request #3348
- Content font size is now configurable, so the log table text can be scaled without browser zoom also scaling the navigation and filter panel. Pull request #3166
- The chart editor alert form now opens in a docked side panel, keeping the preview chart in view while tuning an alert. Pull request #3222
- The external search API no longer drops rows when startTime includes milliseconds, and the maxResults cap is configurable. Pull request #3381
Why it matters
This week brings two new release lines with heatmap modes for builder charts and PromQL, plus a CLI with stacked area charts. Dashboard filters took a step forward with adhoc filter creation and editing, and the API now avoids a public default session secret.
Changelog entry
- App 2.41.0: series mode for builder heatmaps, PromQL series-mode heatmaps, and minimum auto granularity on PromQL sources. Release
- API 2.41.0: no fallback to the public session secret; a random secret is generated and reported once at startup when authentication is enabled. Release
- CLI 0.7.0: stacked line (stacked area) time-series display type. Release
- Support heatmaps on Prometheus histograms with a new histogram mode. Pull request #3332
- Distribution heatmaps calculated from PromQL values with linear or log-spaced buckets. Pull request #3326
- Broadcast adhoc dashboard filter conditions to builder tiles, SQL tiles and PromQL. Pull request #3366
- Stream filter keys and value results to the UI instead of waiting for all values. Pull request #3348
- Content font size is configurable. Pull request #3166
- Edit chart alerts in a docked side panel so the preview chart stays visible. Pull request #3222
- Fix v2 search dropping rows after a sub-second startTime and make the maxResults cap configurable. Pull request #3381
- Wrap long values in release marker tooltips. Pull request #3391
- Fix multi-column ORDER BY without an explicit direction and table header sort toggling. Pull request #3371
Hyperdx app 2.41.0 and API 2.41.0 are out: builder heatmaps get a series mode, PromQL supports series-mode heatmaps and minimum auto granularity, and the API no longer falls back to the public session secret. CLI 0.7.0 adds stacked area charts.
New releases are out for Hyperdx. The app at 2.41.0 adds a series mode to builder heatmaps and PromQL series-mode heatmaps, plus a minimum auto granularity setting for PromQL sources. The API at 2.41.0 no longer falls back to the public session secret: if the secret is unset, it generates a random one and reports it once at startup when authentication is enabled. The CLI at 0.7.0 adds a stacked line time-series display type. On the dashboard side, adhoc filter conditions now broadcast to builder tiles, SQL tiles and PromQL expressions.
Week of October 5, 2026
What shipped
- Fixed how integration test output appears in GitHub Actions logs by running tests directly rather than through nx. Pull request #3346
- Shared wall-clock timeout handling across MCP tools through a common runWithTimeout helper. Pull request #3346
- Fixed metadata Map-key caching so source discovery and AI field discovery no longer reuse each other's limited results. Pull request #3346
- Added a promql variable reference format that becomes the default when no other format is given. Pull request #3346
- Added PromQL heatmaps where the y-axis shows series. Pull request #3346
- Heatmaps now respect an explicitly chosen granularity and the metric source minimum granularity setting. Pull request #3346
- Dashboard variables and filters can now be configured as single-select. Pull request #3346
- Added stacked line charts as a display type. Pull request #3346
- Sessions are no longer signed with the public default secret when no session secret is configured. Pull request #3346
Why it matters
This week adds more control over dashboards and charts, including PromQL heatmaps, stacked lines and single-select filters, plus a new PromQL variable format. It also fixes several reliability issues: the public default session secret, MCP tool timeouts, and metadata caching that could return the wrong keys. Together these changes make charts and queries behave more predictably for anyone running HyperDX.
Changelog entry
- Fixed integration test output formatting in GitHub Actions logs. Pull request #3346
- MCP tools now share wall-clock timeout handling. Pull request #3346
- Metadata Map-key caches are isolated by limit so discovery tools do not reuse limited results. Pull request #3346
- Added a promql variable format, now the default when none is specified. Pull request #3346
- Added PromQL heatmaps with the y-axis as series, also available through the API and MCP. Pull request #3346
- Heatmaps respect explicit granularity and the metric source minimum granularity. Pull request #3346
- Dashboard variables and filters can be configured as single-select. Pull request #3346
- Added stacked line charts. Pull request #3346
- Auto-detect a PromQL source during onboarding and provision one for local dev. Pull request #3346
- Sessions are no longer signed with the public default secret when no session secret is set. Pull request #3346
- Fixed bare search terms against FixedString columns by casting before hasToken. Pull request #3346
This week: PromQL heatmaps, stacked line charts, single-select dashboard filters, heatmap granularity fixes, a new promql variable format, 30s MCP metric budgets, and sessions no longer signed with the public default secret.
This week the project gained PromQL heatmaps in series mode, stacked line charts, single-select dashboard filters, and a new promql variable reference format. Heatmaps now respect an explicitly chosen granularity and the minimum granularity set on a metric source. Reliability work also landed: session cookies are no longer signed with the public default secret when no secret is configured, MCP metric tools moved to a 30 second budget, and metadata Map-key caching was fixed so source discovery and field discovery stop reusing each other's limited results.
Week of September 28, 2026
What shipped
- @hyperdx/app@2.40.0 adds support for macros in PromQL charts, plotting multiple PromQL expressions on one chart, and per-source floor for auto granularity in metric sources. Release
- @hyperdx/otel-collector@2.40.0 compiles in statsdreceiver to ingest StatsD/DogStatsD metrics directly without a separate bridge. Release
- PromQL pie charts and bar charts are now supported for both range queries with reducers and instant queries. Pull request #3253
- Table visualizations based on PromQL sources now display every returned sample as a row with columns for all associated labels. Pull request #3238
- Trace waterfall view now surfaces total duration and span count for the entire trace at a glance. Pull request #3051
- Every ClickHouse query is now tagged with a
log_commentandquery_idto point back to the dashboard tile, saved search, alert, or tool that issued it. Pull request #3155 - PromQL number charts now support sparklines for range queries aligned to the dashboard's time range and granularity. Pull request #3178
- Grafana-compatible PromQL macros $interval and $
rate_intervalare now supported in PromQL expressions. Pull request #3244 - Variables can now be substituted in markdown tiles with automatic escaping for markdown syntax. Pull request #3246
- Chart editor now filters sources by language and display type compatibility, and auto-selects the first allowed source. Pull request #3185
Why it matters
PromQL support expanded significantly with new chart types and macro substitution. Trace debugging improved with waterfall view enhancements and query attribution. The OpenTelemetry collector now handles StatsD metrics natively, reducing infrastructure complexity.
Changelog entry
- OTel collector compiles StatsD receiver for direct metric ingestion Release
- PromQL pie and bar charts support range and instant queries with legend templates Pull request #3253
- PromQL table charts display every sample with columns for all labels Pull request #3238
- PromQL number charts support sparklines from range queries Pull request #3178
- Trace waterfall view shows total duration and span count Pull request #3051
- Queries tagged with
log_commentandquery_idto track origin (dashboard, alert, search, MCP tool) Pull request #3155 - Variables in markdown tiles with automatic markdown escape Pull request #3246
- Chart editor filters sources by config and display type compatibility Pull request #3185
- Row details panel applies source query settings to row lookup Pull request #3234
- Row details panel can show MATERIALIZED and ALIAS columns Pull request #3231
- Settings clause parsing no longer splits on settings inside string literals Pull request #3216
Week of September 21, 2026
What shipped
- Dashboards list redesigned with tabs for All/Favorites/My dashboards, multi-select tag filtering, and sort by last updated. Pull request #3199
- Trace event side panel now has a dedicated Trace logs tab to view logs without hunting through the waterfall. Pull request #3127
- PromQL charts can now plot multiple expressions on a single chart. Pull request #3146
- PromQL number tiles added with support for both instant and range query results. Pull request #3169
- OpenTelemetry collector now includes StatsD/DogStatsD receiver for direct metric ingestion. Pull request #3100
- Number, bar, and pie tiles now pulse during dashboard refresh instead of showing loading state. Pull request #3211
- Heatmap tiles keep current display and pulse during dashboard refresh. Pull request #3221
- Table tiles retain rows on screen and pulse during dashboard refresh. Pull request #3212
- Y-axis ticks on charts now maintain even spacing when Recharts thins them for overlap. Pull request #3172
- Auto Granularity now respects per-source floor to avoid buckets smaller than metric scrape intervals. Pull request #3149
Why it matters
Dashboards are faster to navigate with better organization and filtering. Traces are easier to debug with dedicated log views. PromQL support expands with multiple expressions and number tiles. Metric ingestion improves with StatsD support and better chart display during refreshes.
Changelog entry
- Dashboards list redesigned with All/Favorites/My dashboards tabs, multi-select tag filtering, and sort by last updated Pull request #3199
- Added Trace logs tab to event side panel to view trace logs separately from waterfall Pull request #3127
- PromQL charts now support plotting multiple expressions on a single chart Pull request #3146
- PromQL number tiles added with instant and range query result support Pull request #3169
- OpenTelemetry collector includes StatsD/DogStatsD receiver for direct metric ingestion Pull request #3100
- Auto Granularity respects per-source floor to prevent buckets smaller than metric scrape intervals Pull request #3149
- Number, bar, and pie tiles pulse during dashboard refresh instead of showing loading state Pull request #3211
- Heatmap tiles keep current display with pulse animation during dashboard refresh Pull request #3221
- Table tiles retain rows on screen with pulse animation during dashboard refresh Pull request #3212
- Gauge and sum metric charts can now group by materialized and alias columns Pull request #3191
- Y-axis ticks maintain even spacing when Recharts thins them for label overlap Pull request #3172
- Chart axis ticks no longer lose decimals after abbreviation to k/m/b/t Pull request #3162
- Search editors no longer leak CSS rules during live tail Pull request #3193
- Expanded and collapsed rows in search results no longer corrupt virtualizer heights Pull request #3171
- Row selection checkboxes in search now reveal on hover instead of always showing Pull request #3164
- External API rate limit can now be overridden via environment variable Pull request #3200
Dashboard navigation gets tabs and tag filters. Trace logs move to their own tab. PromQL adds multi-expression charts and number tiles. StatsD ingestion now supported.
This week brings quality-of-life improvements across the board: dashboards are now organized with tabs and filterable by tags, making navigation faster. Traces get a dedicated log tab to cut down debugging time. PromQL support expands with the ability to plot multiple expressions on one chart and use number tiles. On the ingestion side, StatsD/DogStatsD metrics can now be sent directly. Chart interactions also improved—tiles pulse during refreshes instead of showing loading states, and axis spacing is now consistent.
Week of September 14, 2026
What shipped
- Search results now support selecting multiple rows to download or copy as CSV. Pull request #3140
- Alerts page now uses server-side pagination and filtering instead of loading all alerts at once. Pull request #3136
- Dashboard tile alerts can now be exported for Terraform import alongside saved-search alerts. Pull request #3084
- ClickHouse and Prometheus versions upgraded, with Prometheus remote API endpoints now used for PromQL. Pull request #3142
- SQL, Lucene, and PromQL query fields now stay fully visible when unfocused instead of collapsing. Pull request #3150
- Duration-formatted Y-axis labels on charts now use compact formatting to prevent clipping. Release
- Service map nodes now color by absolute error-rate thresholds instead of relative to the highest error. Pull request #3125
- Metric dashboard tiles can now be read and written through the external API. Pull request #3143
- Tags endpoint now filters results by resource type so only relevant tags appear in filters. Pull request #3139
- Added token encryption service for storing sensitive credentials like Slack bot tokens securely. Pull request #3112
Why it matters
Multi-select downloads and server-side alert pagination improve usability at scale. ClickHouse and Prometheus upgrades keep the platform current with dependencies. The UI improvements—expanded query fields, better axis formatting, absolute error thresholds—make charts and filters more reliable.
Changelog entry
- Removed redundant WHERE label and / keycap overlay from query bar; / and s still focus the input Pull request #3152
- Fixed duration axis tick formatter in CLI charts to use compact format consistently Pull request #3158
- Duration-formatted Y-axis labels now use compact formatter to stay within chart's width budget Release
- Fixed Docker build to avoid clobbering base image /usr/lib; enabled image builds on main branch Pull request #3157
- Backfilled alert displayName and tags from referenced saved search or dashboard on API startup Release
- Added dashboard tile alerts to Terraform import exports Release
- Added text indexes for seeding trace schema; compiled spanmetricsconnector for custom OTEL pipelines Release
- Added multi-select support for search rows with CSV download and copy Pull request #3140
- SQL, Lucene, and PromQL query fields now remain fully visible when unfocused Pull request #3150
- Tags endpoint now filters results by resource type Pull request #3139
- Upgraded ClickHouse and Prometheus versions; switched to Prometheus remote API endpoints Pull request #3142
- Alerts page now supports server-side pagination and filtering Pull request #3136
- List alerts API now supports server-side pagination and filtering by tags, creator, source, and search Pull request #3134
- Incident.io webhooks now sent with correct body format that includes required fields Pull request #3144
- Metric dashboards can now be managed as code through the external API Pull request #3143
- Fixed onboarding data probe to escape table and database names in SQL Pull request #3128
- Fixed row side panel for backticked aliases and hyphenated columns Pull request #3115
- Service map nodes now color by absolute error-rate thresholds instead of relative normalization Pull request #3125
- Dashboard tile alerts now exportable for Terraform import Pull request #3084
- Saved-search filters now applied to alert sample query results Pull request #3106
This week: multi-select downloads, alert pagination, ClickHouse/Prometheus upgrades, better query field visibility, improved chart formatting, and metric dashboard API support.
Shipped: multi-row search downloads (CSV format), server-side alert pagination and filtering, dashboard tile alert exports for Terraform, upgraded ClickHouse/Prometheus with remote API endpoints, persistent query field visibility, compact duration axis labels, absolute error-rate thresholds on service maps, metric dashboard API support, and encrypted token storage for integrations.