CLI telemetry preferences
Inspect and control anonymous CLI usage telemetry preferences
The CLI provides telemetry preferences and a staged telemetry core. Normal runs
do not collect or send events yet. To audit locally, set
HYPEQUERY_TELEMETRY_DEBUG=1: valid event envelopes are printed to stderr and
nothing is sent. Maintainers can set HYPEQUERY_TELEMETRY_URL to a staging or
self-hosted first-party ingest endpoint. Opt-outs apply in both modes.
hypequery telemetry status
hypequery telemetry disable
hypequery telemetry enablestatus (also the default action) prints the effective preference, the setting
that determined it, and a random install ID. Saving enable does not override
environment variables or the global flag.
To disable for one invocation, pass --no-telemetry before or after the command,
or set DO_NOT_TRACK=1 or HYPEQUERY_TELEMETRY_DISABLED=1. These take precedence
over saved preferences. Test environments and Hypequery's own CI also disable
telemetry; customer CI follows the saved preference, which defaults to enabled.
Settings are stored in telemetry.json in the CLI config directory. They contain
the preference, a random install ID, a schema version, and an optional notice
timestamp reserved for the first-run notice. The install ID is generated randomly,
not from hardware, a username, or a hostname. Corrupt or unavailable settings
disable telemetry for that run. A failed explicit save reports an error so you
can use an environment variable or the flag instead.
See the CLI reference for precedence and config-directory details. The planned event reference is below; data-handling policy will be documented before default collection is released.
Staging and debugging
HYPEQUERY_TELEMETRY_DEBUG=1 hypequery --version
HYPEQUERY_TELEMETRY_DEBUG=1 hypequery init --no-telemetry # Prints no eventsDebug output contains the exact validated event envelopes, including their
timestamps and the transport-owned $process_person_profile: false and
$geoip_disable: true properties. Invalid events are dropped with a fixed
diagnostic that does not echo rejected input. All telemetry output uses stderr.
Staging delivery uses one in-memory batch per invocation, a built-in Node HTTP(S) request, a 40 ms flush timeout, and a 64 KiB / 32-event cap. Unsent events are dropped; there is no retry process or disk queue. Only HTTPS endpoints and HTTP loopback test endpoints are accepted; URLs with credentials, query strings or fragments are rejected, and redirects are refused. The first-party proxy holds the real PostHog API key and must remove client IP information before forwarding.
HTTP 410 from the proxy disables the current CLI version in local settings,
best-effort. Later versions remain eligible; telemetry status reports
kill_switch when it applies. A read-only config cannot persist the response.
The disclosure appears once on enabled interactive runs, after a successful stderr write. It is suppressed in CI, on non-TTY streams, during preference commands, and under MCP. Before default collection is enabled, this disclosure appears only when the staged core is activated as described above.
Known command failures reach the bounded exit flush. Dev shutdown flushes after runtime teardown, and SIGINT/SIGTERM endings are classified as interrupted. Crash events use Node's exception monitor so native fatal-error output and exit behavior are preserved; network delivery during a fatal crash is best-effort.
In staged mode, init adds onboarding progress to its single command completion
event: the furthest stage reached, the failing stage on errors, connection results,
scaffold and authentication choices, bucketed schema/dataset counts, and whether
configuration files were created, updated, unchanged, or skipped. Declining setup
is recorded as cancellation. Connection retries share the same invocation and
produce one completion event. Table selections are recorded only as a selection
mode; table names, connection details, and output paths are never included.
Generation commands report bucketed counts from the schema they already
introspect. Type fallbacks are reduced to built-in family names (or unknown),
including unsupported fragments inside supported wrappers. Dataset generation
records write/check/diff modes, read-only check outcomes, and counts of known
tenant-scaffolding warnings. Manifest generation counts already loaded registry
entries and generated routes. Schema names, full type expressions, warning text,
and manifest contents are excluded, and telemetry adds no database queries.
Event reference
Schema version: 1. Default collection is inactive. Debug or explicitly configured ingest mode emits command/crash lifecycle events, init and generation metrics, and dev/MCP session events. Cloud command and project-feature instrumentation follows separately.
Unknown events, properties, record keys, flag names, enum values, and malformed identities or versions are rejected. No argument values, paths, names, SQL, messages, stacks, hosts, tenants, or credentials are permitted.
Counts use inclusive integer ranges: 0, 1, 2–5, 6–20, 21–100, and 101 or more. Duration and nonzero size buckets include their lower bound and exclude their upper bound. Raw counts, durations, ports and byte sizes are never sent.
Common properties
Every event includes these fields. Unknown environment details use the documented unknown value; unavailable optional fields are omitted.
| Property | Presence | Allowed values | Question answered |
|---|---|---|---|
schema_version | Required | 1 | Which event contract does this payload use? |
cli_version | Required | Exact major.minor.patch (1–3 digits per component); optional canary/alpha/beta/rc/dev/snapshot suffix with 1–14 digit numeric components, or unknown; at most 80 characters | Which CLI releases are in use? |
node_version | Required | Numeric major.minor (1–3 digits per component), or unknown | Which Node major.minor releases need support? |
os | Required | darwin, linux, win32, aix, freebsd, openbsd, sunos, android, unknown | Which operating systems need support? |
arch | Required | arm, arm64, ia32, x64, loong64, mips, mipsel, ppc, ppc64, riscv64, s390, s390x, unknown | Which CPU architectures need support? |
is_ci | Required | true, false | How much usage happens in CI? |
ci_name | Required | none, github_actions, gitlab, circleci, buildkite, jenkins, vercel, netlify, azure, bitbucket, teamcity, travis, codebuild, unknown | Which CI providers need support? |
is_tty | Required | true, false | How much usage is interactive? |
is_docker | Required | true, false | How much usage runs in containers? |
is_wsl | Required | true, false | How much usage runs in WSL? |
package_manager | Required | npm, pnpm, yarn, bun, unknown | Which package managers need support? |
package_manager_major | Optional | Numeric major (1–3 digits), or unknown | Which package-manager major releases need support? |
invoked_via | Required | npx, pnpm_dlx, bunx, global, local_bin, unknown | How is the CLI invoked? |
install_id | Required | Random UUID v4 (36 characters) | Count active installations without hardware or account identifiers. |
project_id | Optional | 64 lowercase hexadecimal characters (salted SHA-256) | Count active projects across installations; omit when no signal exists. |
session_id | Required | Random UUID v4 (36 characters) | Group events from the same invocation. |
is_first_run | Required | true, false | How many installations are new? |
hypequery_packages | Optional | Object with only the documented keys below | Which toolkit packages and exact versions are adopted? No project dependencies or names. |
hypequery_packages.@hypequery/clickhouse | Optional | Exact major.minor.patch (1–3 digits per component); optional canary/alpha/beta/rc/dev/snapshot suffix with 1–14 digit numeric components, or unknown; at most 80 characters | Installed exact release version, or unknown. |
hypequery_packages.@hypequery/datasets | Optional | Exact major.minor.patch (1–3 digits per component); optional canary/alpha/beta/rc/dev/snapshot suffix with 1–14 digit numeric components, or unknown; at most 80 characters | Installed exact release version, or unknown. |
hypequery_packages.@hypequery/serve | Optional | Exact major.minor.patch (1–3 digits per component); optional canary/alpha/beta/rc/dev/snapshot suffix with 1–14 digit numeric components, or unknown; at most 80 characters | Installed exact release version, or unknown. |
hypequery_packages.@hypequery/react | Optional | Exact major.minor.patch (1–3 digits per component); optional canary/alpha/beta/rc/dev/snapshot suffix with 1–14 digit numeric components, or unknown; at most 80 characters | Installed exact release version, or unknown. |
hypequery_packages.@hypequery/cli | Optional | Exact major.minor.patch (1–3 digits per component); optional canary/alpha/beta/rc/dev/snapshot suffix with 1–14 digit numeric components, or unknown; at most 80 characters | Installed exact release version, or unknown. |
hypequery_packages.@hypequery/mcp | Optional | Exact major.minor.patch (1–3 digits per component); optional canary/alpha/beta/rc/dev/snapshot suffix with 1–14 digit numeric components, or unknown; at most 80 characters | Installed exact release version, or unknown. |
hypequery_packages.@hypequery/protocol | Optional | Exact major.minor.patch (1–3 digits per component); optional canary/alpha/beta/rc/dev/snapshot suffix with 1–14 digit numeric components, or unknown; at most 80 characters | Installed exact release version, or unknown. |
hypequery_packages.@hypequery/protocol-conformance | Optional | Exact major.minor.patch (1–3 digits per component); optional canary/alpha/beta/rc/dev/snapshot suffix with 1–14 digit numeric components, or unknown; at most 80 characters | Installed exact release version, or unknown. |
hypequery_packages.@hypequery/deployment | Optional | Exact major.minor.patch (1–3 digits per component); optional canary/alpha/beta/rc/dev/snapshot suffix with 1–14 digit numeric components, or unknown; at most 80 characters | Installed exact release version, or unknown. |
database | Required | clickhouse, chdb, bigquery, unknown | Which database backends merit investment? |
cli_command_completed
Measure command adoption, onboarding drop-off and release reliability. Exactly once per invocation; never emit for telemetry disable.
| Property | Presence | Allowed values | Question answered |
|---|---|---|---|
command | Required | init, dev, mcp, generate, generate:types, generate:datasets, generate:manifest, login, logout, deploy, deployment:build, deployment:validate, deployment:release, deployment:submit, deployment:status, pull, diff, help, version, telemetry, unknown | Which known command was requested? Raw unknown input is never retained. |
flags_used | Required | Unique list of: --path, --style, --database, --chdb-path, --auth, --all-tables, --tables, --exclude-tables, --no-example, --no-interactive, --force, --skip-connection, --port, --hostname, --no-watch, --no-cache, --cache, --redis-url, --open, --cors, --quiet, --tenant, --self-test, --url, --output, --tenant-column, --check, --diff, --cloud-url, --environment, --bundle-output, --release-output, --project, --no-source, --endpoint, --replace-restored, --mcp-config, --release, --no-telemetry, --help, --version | Which documented flags are adopted? Canonical names only, never values. |
outcome | Required | success, failure, cancelled, interrupted | Which commands fail, succeed or are aborted? |
duration_bucket | Required | <100ms, 100ms-1s, 1s-10s, 10s-1m, 1m-10m, 10m-1h, 1h+ | Which commands need performance work? |
error_code | Optional | unknown, connection_failed, auth_failed, entrypoint_not_found, compile_error, prompt_cancelled, validation_failed, cloud_network_error, cloud_rejected, not_installed, engine_error, no_tables, no_datasets, tenant_required, load_error, output_missing, output_out_of_date, unknown_command, unknown_option | Which known failures need reliability work? |
cli_command_completed: init
Allowed flags_used: --path, --style, --database, --chdb-path, --auth, --all-tables, --tables, --exclude-tables, --no-example, --no-interactive, --force, --skip-connection, --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
| Property | Presence | Allowed values | Question answered |
|---|---|---|---|
stage_reached | Optional | started, database_selected, connection_tested, style_selected, files_written, dependencies_installed, completed | Where does onboarding stop? |
failed_stage | Optional | started, database_selected, connection_tested, style_selected, files_written, dependencies_installed, completed | Which onboarding stage needs reliability work? |
style | Optional | queries, datasets | Which scaffold style is adopted? |
auth | Optional | none, context | Is context authentication adopted? |
interactive | Optional | true, false | How much onboarding is interactive? |
force | Optional | true, false | How often are scaffold files overwritten? |
example | Optional | true, false | Are example queries requested? |
skip_connection | Optional | true, false | Are offline scaffolds used? |
connection_result | Optional | ok, failed, skipped | Where does connection setup fail? |
chdb_failure_reason | Optional | not_installed, engine_error | Does embedded onboarding fail at installation or execution? |
table_count_bucket | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | How large are onboarding schemas? |
datasets_generated_bucket | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | How many datasets are scaffolded? |
table_selection | Optional | all, list, exclude, prompt | Which table-selection workflows are used? |
env_file | Optional | created, updated, unchanged, skipped | Does onboarding create or update environment config? |
gitignore | Optional | created, updated, unchanged, skipped | Does onboarding create or update ignore config? |
package_json_present | Optional | true, false | Does onboarding start in an existing project? |
analytics_directory_present | Optional | true, false | Does onboarding reuse an analytics directory? |
cli_command_completed: dev
Allowed flags_used: --port, --hostname, --no-watch, --no-cache, --cache, --redis-url, --open, --cors, --path, --quiet, --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
No additional properties.
cli_command_completed: mcp
Allowed flags_used: --path, --tenant, --self-test, --url, --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
| Property | Presence | Allowed values | Question answered |
|---|---|---|---|
mode | Optional | self_test, serve | Is MCP used for validation or serving? |
hosted | Optional | true, false | Is MCP self-test targeting a hosted endpoint? |
dataset_count_bucket | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | How many datasets are exposed to agents? |
tenant_used | Optional | true, false | Is a trusted tenant supplied? Never its value. |
tenant_dataset_count_bucket | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | How many datasets require tenant scope? |
cli_command_completed: generate
Allowed flags_used: --output, --path, --tables, --database, --chdb-path, --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
| Property | Presence | Allowed values | Question answered |
|---|---|---|---|
chdb_path_given | Optional | true, false | Is persistent embedded ClickHouse adopted? |
tables_filter_used | Optional | true, false | Are targeted type-generation workflows used? |
table_count_bucket | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | How large are introspected schemas? |
column_count_bucket | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | How many columns need type generation? |
unsupported_type_count_bucket | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | How often does type mapping need improvement? |
unsupported_type_families | Optional | Unique list of: AggregateFunction, SimpleAggregateFunction, Variant, Dynamic, Object, JSON, Nested, Tuple, Map, Array, Nullable, LowCardinality, Decimal, Enum, DateTime, DateTime64, unknown | Which built-in type families need mapping? Never full type expressions. |
custom_output | Optional | true, false | Was a custom output location supplied? Never the path. |
cli_command_completed: generate:types
Allowed flags_used: --output, --path, --tables, --database, --chdb-path, --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
| Property | Presence | Allowed values | Question answered |
|---|---|---|---|
chdb_path_given | Optional | true, false | Is persistent embedded ClickHouse adopted? |
tables_filter_used | Optional | true, false | Are targeted type-generation workflows used? |
table_count_bucket | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | How large are introspected schemas? |
column_count_bucket | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | How many columns need type generation? |
unsupported_type_count_bucket | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | How often does type mapping need improvement? |
unsupported_type_families | Optional | Unique list of: AggregateFunction, SimpleAggregateFunction, Variant, Dynamic, Object, JSON, Nested, Tuple, Map, Array, Nullable, LowCardinality, Decimal, Enum, DateTime, DateTime64, unknown | Which built-in type families need mapping? Never full type expressions. |
custom_output | Optional | true, false | Was a custom output location supplied? Never the path. |
cli_command_completed: generate:datasets
Allowed flags_used: --output, --path, --tables, --exclude-tables, --tenant-column, --force, --check, --diff, --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
| Property | Presence | Allowed values | Question answered |
|---|---|---|---|
mode | Optional | write, check, diff | Are datasets generated locally or checked in CI? |
check_result | Optional | up_to_date, out_of_date, missing | Does CI find stale dataset definitions? |
force | Optional | true, false | Are existing definitions replaced? |
tenant_column_used | Optional | true, false | Is tenant isolation explicitly scaffolded? Never the column. |
tables_filter_used | Optional | true, false | Is include filtering adopted? |
exclude_filter_used | Optional | true, false | Is exclude filtering adopted? |
datasets_generated_bucket | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | How many dataset definitions are generated? |
warning_counts | Optional | Object with only the documented keys below | Which tenant-scaffolding warnings need improvement? |
warning_counts.tenant-key-candidate | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed warning occurrences. |
warning_counts.tenant-column-missing | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed warning occurrences. |
cli_command_completed: generate:manifest
Allowed flags_used: --output, --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
| Property | Presence | Allowed values | Question answered |
|---|---|---|---|
query_count_bucket | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | How much query-based React integration is adopted? |
endpoint_count_bucket | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | How much endpoint-based React integration is adopted? |
custom_output | Optional | true, false | Was a custom output location supplied? Never the path. |
cli_command_completed: login
Allowed flags_used: --cloud-url, --environment, --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
No additional properties.
cli_command_completed: logout
Allowed flags_used: --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
No additional properties.
cli_command_completed: deploy
Allowed flags_used: --bundle-output, --release-output, --project, --environment, --no-source, --endpoint, --replace-restored, --mcp-config, --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
No additional properties.
cli_command_completed: deployment:build
Allowed flags_used: --bundle-output, --no-source, --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
No additional properties.
cli_command_completed: deployment:validate
Allowed flags_used: --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
No additional properties.
cli_command_completed: deployment:release
Allowed flags_used: --project, --environment, --output, --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
No additional properties.
cli_command_completed: deployment:submit
Allowed flags_used: --release, --endpoint, --replace-restored, --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
No additional properties.
cli_command_completed: deployment:status
Allowed flags_used: --project, --environment, --endpoint, --mcp-config, --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
No additional properties.
cli_command_completed: pull
Allowed flags_used: --output, --project, --environment, --endpoint, --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
No additional properties.
cli_command_completed: diff
Allowed flags_used: --project, --environment, --endpoint, --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
No additional properties.
cli_command_completed: version
Allowed flags_used: --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
No additional properties.
cli_command_completed: telemetry
Allowed flags_used: --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
No additional properties.
cli_command_completed: unknown
Allowed flags_used: --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
No additional properties.
cli_command_completed: help
Allowed flags_used: --no-telemetry, --help, --version. The two type-generation aliases remain distinct command values.
| Property | Presence | Allowed values | Question answered |
|---|---|---|---|
help_topic | Optional | root, init, dev, mcp, generate, generate:types, generate:datasets, generate:manifest, login, logout, deploy, deployment:build, deployment:validate, deployment:release, deployment:submit, deployment:status, pull, diff, help, version, telemetry, unknown | Which command help needs improvement? Unknown input becomes unknown. |
cli_session_started
Measure development-server and MCP session configuration; never emit on every reload or request.
| Property | Presence | Allowed values | Question answered |
|---|---|---|---|
command | Required | dev, mcp | Which long-running command starts? |
cli_session_started: dev
| Property | Presence | Allowed values | Question answered |
|---|---|---|---|
entry_type | Required | hypequery.ts, api.ts, queries.ts, explicit_file, unknown | Which entrypoint convention is used? Never the path. |
watch | Required | true, false | Is file watching used? |
cache_provider | Required | none, memory, redis, unknown | Which cache providers are adopted? |
cors | Required | true, false | Is cross-origin development adopted? |
open | Required | true, false | Is the browser opened automatically? |
custom_port | Required | true, false | Is a non-default port selected? Never its number or hostname. |
quiet | Required | true, false | Is quiet startup adopted? |
cli_session_started: mcp
| Property | Presence | Allowed values | Question answered |
|---|---|---|---|
entry_type | Required | hypequery.ts, api.ts, queries.ts, explicit_file, unknown | Which entrypoint convention is used? Never the path. |
mode | Required | self_test, serve | Is MCP serving or validating? |
dataset_count_bucket | Required | 0, 1, 2-5, 6-20, 21-100, 101+ | How many datasets are exposed to agents? |
tenant_used | Required | true, false | Is tenant scope supplied? Never its value. |
tenant_dataset_count_bucket | Required | 0, 1, 2-5, 6-20, 21-100, 101+ | How many exposed datasets require tenant scope? |
cli_session_ended
Measure session duration and aggregated usage on shutdown, without individual requests or tool arguments.
| Property | Presence | Allowed values | Question answered |
|---|---|---|---|
command | Required | dev, mcp | Which long-running command ends? |
outcome | Required | success, failure, cancelled, interrupted | How did the session end? |
duration_bucket | Required | <100ms, 100ms-1s, 1s-10s, 10s-1m, 1m-10m, 10m-1h, 1h+ | How long are development and agent sessions? |
cli_session_ended: dev
| Property | Presence | Allowed values | Question answered |
|---|---|---|---|
reload_count_bucket | Required | 0, 1, 2-5, 6-20, 21-100, 101+ | How heavily is hot reload used? |
reload_error_count_bucket | Required | 0, 1, 2-5, 6-20, 21-100, 101+ | How often does hot reload fail? |
load_failures | Required | Object with only the documented keys below | Which load failures need reliability work? |
load_failures.unknown | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.connection_failed | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.auth_failed | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.entrypoint_not_found | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.compile_error | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.prompt_cancelled | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.validation_failed | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.cloud_network_error | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.cloud_rejected | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.not_installed | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.engine_error | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.no_tables | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.no_datasets | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.tenant_required | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.load_error | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.output_missing | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.output_out_of_date | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.unknown_command | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
load_failures.unknown_option | Optional | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed failures by stable code. |
cli_session_ended: mcp
| Property | Presence | Allowed values | Question answered |
|---|---|---|---|
tool_call_counts | Required | Object with only the documented keys below | Which MCP tool kinds are adopted? No names or arguments. |
tool_call_counts.list | Required | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed calls by built-in tool kind. |
tool_call_counts.describe | Required | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed calls by built-in tool kind. |
tool_call_counts.query | Required | 0, 1, 2-5, 6-20, 21-100, 101+ | Bucketed calls by built-in tool kind. |
error_count_bucket | Required | 0, 1, 2-5, 6-20, 21-100, 101+ | How often do MCP sessions encounter errors? |
cli_crash
Measure uncaught failures by stable class and code, with no messages or stacks.
| Property | Presence | Allowed values | Question answered |
|---|---|---|---|
command | Required | init, dev, mcp, generate, generate:types, generate:datasets, generate:manifest, login, logout, deploy, deployment:build, deployment:validate, deployment:release, deployment:submit, deployment:status, pull, diff, help, version, telemetry, unknown | Which command crashed? |
exception_class | Required | Error, TypeError, RangeError, SyntaxError, ReferenceError, URIError, EvalError, AggregateError, unknown | Which built-in exception classes need investigation? |
error_code | Required | unknown, connection_failed, auth_failed, entrypoint_not_found, compile_error, prompt_cancelled, validation_failed, cloud_network_error, cloud_rejected, not_installed, engine_error, no_tables, no_datasets, tenant_required, load_error, output_missing, output_out_of_date, unknown_command, unknown_option | Which known failure caused the crash? |
MCP client identity/version is excluded from v1 pending the privacy decision. Auth strategies are recorded only as none/configured/unknown, since user-supplied functions have no safe built-in type identifier.
Cache scopes are classified only as unscoped/tenant/explicit/unknown from configuration presence. Explicit scopes are user content and are never read into a payload.
Project IDs follow HQ-233: the CLI uses the shared global salt, then the ingest proxy re-keys the digest with a server-side HMAC secret before storage. A hash is a pseudonymous project signal, not proof of anonymity. Cloud accounts are never linked to install IDs.