@hyperdx/otel-collector@2.27.0

@hyperdx/otel-collector@2.27.0

@hyperdx/common-utils@0.19.1

Patch Changes

• 84117a7: fix: support CAST() form in KV items column expression parsing for direct_read optimization
• 51abe98: fix: Event Patterns and other CTE-using queries now correctly detect Date-typed partition columns and wrap them in toDate(), fixing "No results found" against sources with a Date partition key (e.g. event_date / EventDate).

@hyperdx/cli@0.4.1

Patch Changes

• f6a1d02: Add support for event patterns in MCP server, reduce code duplication

• 253cf5b: Fix CLI version flag reporting hardcoded 0.1.0 instead of the actual package version

• 4104364: feat: support multiple teams and kubectx-style team switching in the CLI

  Adds three new commands for users that belong to multiple teams (HyperDX Cloud /
  EE):

  • hdx team list — list every team the authenticated user belongs to, marking
    the active one
  • hdx team current — print the currently active team
  • hdx team use <name-or-id> — switch the active team (matched by team ID or
    case-insensitive name)

  The active team is persisted to ~/.config/hyperdx/cli/session.json so the
  choice survives across CLI invocations, and the CLI now sends an x-hdx-team
  header on every API and ClickHouse-proxy request so the server scopes data to
  the chosen team. hdx auth status also surfaces the active team.

  On single-team OSS deployments these commands are effectively no-ops.

@hyperdx/app@2.27.0

Patch Changes

• Updated dependencies [f5ae006]
• Updated dependencies [fbe5a9a]
  • @hyperdx/api@2.27.0

@hyperdx/api@2.27.0

Minor Changes

• fbe5a9a: feat: Add POST /api/v2/search endpoint for querying raw log and trace rows programmatically

Patch Changes

• f5ae006: refactor(mcp): split hyperdx_query into 5 display-type-specific tools

  Replace the monolithic hyperdx_query tool with five narrow tools:

  • hyperdx_timeseries (line + stacked_bar)
  • hyperdx_table (table + number + pie, with shape auto-upgrade)
  • hyperdx_search (raw event browsing)
  • hyperdx_event_patterns (Drain pattern mining)
  • hyperdx_sql (raw ClickHouse SQL)

  Each tool's schema contains only its relevant parameters — no displayType
  discriminator, no fields from other modes, no conditional required fields.
  hyperdx_query is removed from the tool surface.

@hyperdx/cli v0.4.1

@hyperdx/cli v0.4.1

Patch Changes

• f6a1d02: Add support for event patterns in MCP server, reduce code duplication

• 253cf5b: Fix CLI version flag reporting hardcoded 0.1.0 instead of the actual package version

• 4104364: feat: support multiple teams and kubectx-style team switching in the CLI

  Adds three new commands for users that belong to multiple teams (HyperDX Cloud /
  EE):

  • hdx team list — list every team the authenticated user belongs to, marking
    the active one
  • hdx team current — print the currently active team
  • hdx team use <name-or-id> — switch the active team (matched by team ID or
    case-insensitive name)

  The active team is persisted to ~/.config/hyperdx/cli/session.json so the
  choice survives across CLI invocations, and the CLI now sends an x-hdx-team
  header on every API and ClickHouse-proxy request so the server scopes data to
  the chosen team. hdx auth status also surfaces the active team.

  On single-team OSS deployments these commands are effectively no-ops.

---

Installation

npm (recommended):

npm install -g @hyperdx/cli

Or run directly with npx:

npx @hyperdx/cli tui -s <your-hyperdx-api-url>

Manual download (standalone binary, no Node.js required):

# macOS Apple Silicon
curl -L https://github.com/hyperdxio/hyperdx/releases/download/cli-v0.4.1/hdx-darwin-arm64 -o hdx
# macOS Intel
curl -L https://github.com/hyperdxio/hyperdx/releases/download/cli-v0.4.1/hdx-darwin-x64 -o hdx
# Linux x64
curl -L https://github.com/hyperdxio/hyperdx/releases/download/cli-v0.4.1/hdx-linux-x64 -o hdx

chmod +x hdx && sudo mv hdx /usr/local/bin/

Usage

hdx auth login -s <your-hyperdx-api-url>
hdx tui

@hyperdx/cli v0.4.0

@hyperdx/cli v0.4.0

Installation

npm (recommended):

npm install -g @hyperdx/cli

Or run directly with npx:

npx @hyperdx/cli tui -s <your-hyperdx-api-url>

Manual download (standalone binary, no Node.js required):

# macOS Apple Silicon
curl -L https://github.com/hyperdxio/hyperdx/releases/download/cli-v0.4.0/hdx-darwin-arm64 -o hdx
# macOS Intel
curl -L https://github.com/hyperdxio/hyperdx/releases/download/cli-v0.4.0/hdx-darwin-x64 -o hdx
# Linux x64
curl -L https://github.com/hyperdxio/hyperdx/releases/download/cli-v0.4.0/hdx-linux-x64 -o hdx

chmod +x hdx && sudo mv hdx /usr/local/bin/

Usage

hdx auth login -s <your-hyperdx-api-url>
hdx tui

@hyperdx/otel-collector@2.24.1

Patch Changes

• 11e1301: feat(otel-collector): tune batch processor defaults for ClickHouse and make
  them configurable

  The bundled OTel Collector config now sets processors.batch.send_batch_size
  to 10000 and timeout to 5s (was upstream defaults of 8192 / 200ms),
  matching the values recommended by the ClickHouse exporter and ClickStack
  docs. The upstream defaults were too aggressive for ClickHouse — they
  produced very small inserts under low load, hurting insert performance and
  inflating the number of MergeTree parts.

  Each setting can also be overridden via environment variables:

  • HYPERDX_OTEL_BATCH_SEND_BATCH_SIZE (default: 10000)
  • HYPERDX_OTEL_BATCH_SEND_BATCH_MAX_SIZE (default: 0, meaning no upper
    bound)
  • HYPERDX_OTEL_BATCH_TIMEOUT (default: 5s)

@hyperdx/common-utils@0.18.1

Patch Changes

• b73f6fc: fix: Prevent duplicate tile IDs in dashboard imports
• 4c23e10: feat: Allow displaying group-by columns on LHS of table
• e2fc25d: feat: Add custom table onClick behavior
• 7665fbe: refactor: Unify section/group into single Group with collapsible/bordered options

@hyperdx/cli@0.4.0

Minor Changes

• 3571bfa: Add hdx query and hdx connections commands for agentic ClickHouse workflows:

  • hdx connections — list ClickHouse connections (id, name, host) for the authenticated team. Supports --json for programmatic consumption.
  • hdx query --connection-id <id> --sql <query> — run raw SQL against a configured ClickHouse connection via the /clickhouse-proxy. Default --format is JSONEachRow (NDJSON); any ClickHouse format is accepted. Exit codes: 0 on success (empty stdout = zero rows), 1 on failure. On error, a lazy connection lookup distinguishes "unknown connection ID" from "bad SQL".
  • hdx query --patterns — post-process the result with the Drain algorithm (@hyperdx/common-utils/dist/drain) and emit one NDJSON pattern object per cluster, sorted by count desc: {"pattern":"<template>","count":<n>,"sample":"<first sample>"}. Use --body-column <name> to cluster a single column's string value; defaults to the whole row JSON-serialized so any SELECT shape works.
  • --help documents the exit-code contract and includes sampling guidance (ORDER BY rand() + selective WHERE warning) for mining over large tables.