:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.
Python ab-testingai-analyticsanalyticscdpdata-warehouseexperimentsfeature-flagsjavascriptproduct-analyticspythonreactsession-replaysurveystypescriptweb-analytics
feat(mcp): give replay vision scanner tools a compact recording filter schema (#125746)
Kim Svatos Dugan8c8bc1bb127d
diff --git a/products/replay_vision/backend/queries/scanner_volume_estimate.py b/products/replay_vision/backend/queries/scanner_volume_estimate.pyindex a22b42a317af..39d7e5727eeb 100644--- a/products/replay_vision/backend/queries/scanner_volume_estimate.py+++ b/products/replay_vision/backend/queries/scanner_volume_estimate.py@@ -105,6 +105,11 @@ class MatchCheckIncomplete(Exception):"""The match check could not reach an answer: every fetched recording was excluded and more remain."""+def _combines_with_or(query: RecordingsQuery) -> bool:+ # Read the way the recordings query builder reads it: every operand but AND, an explicit null included, is OR.+ return query.operand != FilterLogicalOperator.AND_++def has_recent_matching_recording(*,team: Team,@@ -134,7 +139,7 @@ def has_recent_matching_recording(raise TimeoutError("The recent-recording match check ran out of time.")return remaining- scan_window_days = budget.window_days(unsampled=query.operand == FilterLogicalOperator.OR_)+ scan_window_days = budget.window_days(unsampled=_combines_with_or(query))windowed = query.model_copy(deep=True)windowed.date_from = (dt.datetime.now(dt.UTC) - dt.timedelta(days=scan_window_days)).isoformat()windowed.date_to = None@@ -197,7 +202,7 @@ def estimate_scanner_session_volume(# Sampling is only sound when every match must pass the sampled events leg; under OR, sessions# matched via unsampled branches (persons, cohorts, console logs) would be multiplied by the# correction. Without sampling the scan is full price, so bound its window instead.- unsampled = query.operand == FilterLogicalOperator.OR_+ unsampled = _combines_with_or(query)sample_factor = None if unsampled else _ESTIMATE_EVENTS_SAMPLE_FACTORscan_window_days = budget.window_days(unsampled=unsampled)diff --git a/products/replay_vision/backend/tests/test_scanner_candidate_query.py b/products/replay_vision/backend/tests/test_scanner_candidate_query.pyindex 6f8b76e03ad4..96d9a403f1e1 100644--- a/products/replay_vision/backend/tests/test_scanner_candidate_query.py+++ b/products/replay_vision/backend/tests/test_scanner_candidate_query.py@@ -511,8 +511,9 @@ class TestScannerCandidateQueryAgainstClickHouse(ClickhouseTestMixin):assert estimate.matched_sessions == 1+ @pytest.mark.parametrize("operand", [FilterLogicalOperator.OR_, None])@pytest.mark.django_db- def test_unsampled_estimate_window_narrows_only_for_previews(self, team) -> None:+ def test_unsampled_estimate_window_narrows_only_for_previews(self, team, operand) -> None:# An OR operand rules out the events SAMPLE, so a preview scans a shorter window instead of# paying full price. A persisted estimate keeps the whole week, because a partial week# inherits whichever weekdays it covered and biases the monthly projection.@@ -525,8 +526,9 @@ class TestScannerCandidateQueryAgainstClickHouse(ClickhouseTestMixin):timestamp=first,properties={"$session_id": "older-match"},)+ # The query builder reads a null operand as OR too, so the estimate must not sample it.query = RecordingsQuery(- operand=FilterLogicalOperator.OR_,+ operand=operand,events=[{"id": "$pageview", "type": "events", "order": 0, "name": "$pageview"}],)diff --git a/products/replay_vision/mcp/tools.yaml b/products/replay_vision/mcp/tools.yamlindex b42f1f47f784..b5b74c1be40a 100644--- a/products/replay_vision/mcp/tools.yaml+++ b/products/replay_vision/mcp/tools.yaml@@ -608,12 +608,16 @@ tools:`_posthogUrl` opens the scanner where they do it. There is nothing to rate yet, so this is a closingsentence for them, not a step for you.exclude_params:+ - query- id- scanner_version- last_swept_at- created_at- created_by- updated_at+ param_overrides:+ query:+ input_schema: ScannerRecordingsFilterSchemarequires_ai_consent: truevision-scanners-dashboard-create:operation: vision_scanners_create_dashboard_create@@ -703,17 +707,23 @@ tools:title: Estimate replay vision scanner volumedescription: >Preview how many observations a proposed scanner would generate, and what they would cost, before saving it.- Send a `query` (`RecordingsQuery` shape; `date_from`/`date_to` are ignored — the estimate uses a fixed- 30-day lookback) plus the proposed `sampling_rate`, `sampling_mode` and `model`, and `scanner_id` when- re-estimating an existing scanner so its own projection isn't double-counted. For an experiment scanner,- pass `experiment` with the same `experiment_id` and `variants` you will save, so the estimate counts only- exposed sessions. Returns `matched_sessions_in_window`, the `window_days` it was measured over,- `estimated_observations_per_month`, `credits_per_observation`, `estimated_credits_per_month`, and- `other_enabled_scanners_monthly_credits` (the org's other enabled scanners' projected spend). Enabled- scanners sweep every 5 minutes, so a broad query can exhaust the budget quickly — run this before- `vision-scanners-create` and compare the *credit* figures (never raw observation counts) against `remaining`- from `vision-quota-get`. These projections are per month while `remaining` covers only the rest of the- current period, so prorate them by the days left between `period_start` and `period_end` before comparing.+ Send a `query` (`RecordingsQuery` shape; `date_from`/`date_to` are ignored, because the estimate counts a+ recent window, `window_days` in the response, and scales it to 30 days) plus the proposed `sampling_rate`,+ `sampling_mode` and `model`, and `scanner_id` when re-estimating an existing scanner so its own projection+ isn't double-counted. For an experiment scanner, pass `experiment` with the same `experiment_id` and+ `variants` you will save, so the estimate counts only exposed sessions. Returns+ `matched_sessions_in_window`, the `window_days` it was measured over, `estimated_observations_per_month`,+ `credits_per_observation`, `estimated_credits_per_month`, and `other_enabled_scanners_monthly_credits` (the+ org's other enabled scanners' projected spend). Enabled scanners sweep every 5 minutes, so a broad query can+ exhaust the budget quickly — run this before `vision-scanners-create` and compare the *credit* figures+ (never raw observation counts) against `remaining` from `vision-quota-get`. These projections are per month+ while `remaining` covers only the rest of the current period, so prorate them by the days left between+ `period_start` and `period_end` before comparing.+ exclude_params:+ - query+ param_overrides:+ query:+ input_schema: ScannerEstimateRecordingsFilterSchemavision-scanners-get:operation: vision_scanners_retrieveenabled: true@@ -1071,21 +1081,26 @@ tools:idempotent: truetitle: Update replay vision scannerdescription: >- Partially update a replay vision scanner by ID. Send only the fields you want to change. Common changes:- toggle `enabled` to pause/resume scanning, adjust `sampling_rate` or `sampling_mode`, tweak the- `scanner_config` prompt or tags, or refine the `query` to watch a different slice of sessions.- `scanner_type` is locked after creation — to change it, delete and recreate. Editing any config field bumps- `scanner_version`; existing observations keep a snapshot of the previous config for reproducibility. If you- widen the `query`, raise `sampling_rate`, loosen `sampling_mode`, or move to a pricier `model`, re-check- projected spend with `vision-scanners-estimate` and remaining budget with `vision-quota-get` first — the- scanner resumes sweeping every 5 minutes at the new, broader scope.+ Partially update a replay vision scanner by ID. Send only the fields you want to change, except `query`: it+ replaces the stored filter as a whole, so read it with `vision-scanners-get` and send it back in full with+ your changes. Common changes: toggle `enabled` to pause/resume scanning, adjust `sampling_rate` or+ `sampling_mode`, tweak the `scanner_config` prompt or tags, or refine the `query` to watch a different slice+ of sessions. `scanner_type` is locked after creation — to change it, delete and recreate. Editing any config+ field bumps `scanner_version`; existing observations keep a snapshot of the previous config for+ reproducibility. If you widen the `query`, raise `sampling_rate`, loosen `sampling_mode`, or move to a+ pricier `model`, re-check projected spend with `vision-scanners-estimate` and remaining budget with+ `vision-quota-get` first — the scanner resumes sweeping every 5 minutes at the new, broader scope.exclude_params:+ - query- id- scanner_version- last_swept_at- created_at- created_by- updated_at+ param_overrides:+ query:+ input_schema: ScannerUpdateRecordingsFilterSchemarequires_ai_consent: truevision-scanners-variants-list:operation: vision_scanners_variants_listdiff --git a/services/mcp/schema/generated-tool-definitions.json b/services/mcp/schema/generated-tool-definitions.jsonindex 2f4a2fe65868..b7a537f2cd6c 100644--- a/services/mcp/schema/generated-tool-definitions.json+++ b/services/mcp/schema/generated-tool-definitions.json@@ -15813,7 +15813,7 @@"requires_ai_consent": true},"vision-scanners-estimate": {- "description": "Preview how many observations a proposed scanner would generate, and what they would cost, before saving it. Send a `query` (`RecordingsQuery` shape; `date_from`/`date_to` are ignored — the estimate uses a fixed 30-day lookback) plus the proposed `sampling_rate`, `sampling_mode` and `model`, and `scanner_id` when re-estimating an existing scanner so its own projection isn't double-counted. For an experiment scanner, pass `experiment` with the same `experiment_id` and `variants` you will save, so the estimate counts only exposed sessions. Returns `matched_sessions_in_window`, the `window_days` it was measured over, `estimated_observations_per_month`, `credits_per_observation`, `estimated_credits_per_month`, and `other_enabled_scanners_monthly_credits` (the org's other enabled scanners' projected spend). Enabled scanners sweep every 5 minutes, so a broad query can exhaust the budget quickly — run this before `vision-scanners-create` and compare the *credit* figures (never raw observation counts) against `remaining` from `vision-quota-get`. These projections are per month while `remaining` covers only the rest of the current period, so prorate them by the days left between `period_start` and `period_end` before comparing.",+ "description": "Preview how many observations a proposed scanner would generate, and what they would cost, before saving it. Send a `query` (`RecordingsQuery` shape; `date_from`/`date_to` are ignored, because the estimate counts a recent window, `window_days` in the response, and scales it to 30 days) plus the proposed `sampling_rate`, `sampling_mode` and `model`, and `scanner_id` when re-estimating an existing scanner so its own projection isn't double-counted. For an experiment scanner, pass `experiment` with the same `experiment_id` and `variants` you will save, so the estimate counts only exposed sessions. Returns `matched_sessions_in_window`, the `window_days` it was measured over, `estimated_observations_per_month`, `credits_per_observation`, `estimated_credits_per_month`, and `other_enabled_scanners_monthly_credits` (the org's other enabled scanners' projected spend). Enabled scanners sweep every 5 minutes, so a broad query can exhaust the budget quickly — run this before `vision-scanners-create` and compare the *credit* figures (never raw observation counts) against `remaining` from `vision-quota-get`. These projections are per month while `remaining` covers only the rest of the current period, so prorate them by the days left between `period_start` and `period_end` before comparing.","category": "Replay vision","feature": "replay_vision","summary": "Estimate replay vision scanner volume",@@ -16057,7 +16057,7 @@}},"vision-scanners-update": {- "description": "Partially update a replay vision scanner by ID. Send only the fields you want to change. Common changes: toggle `enabled` to pause/resume scanning, adjust `sampling_rate` or `sampling_mode`, tweak the `scanner_config` prompt or tags, or refine the `query` to watch a different slice of sessions. `scanner_type` is locked after creation — to change it, delete and recreate. Editing any config field bumps `scanner_version`; existing observations keep a snapshot of the previous config for reproducibility. If you widen the `query`, raise `sampling_rate`, loosen `sampling_mode`, or move to a pricier `model`, re-check projected spend with `vision-scanners-estimate` and remaining budget with `vision-quota-get` first — the scanner resumes sweeping every 5 minutes at the new, broader scope.",+ "description": "Partially update a replay vision scanner by ID. Send only the fields you want to change, except `query`: it replaces the stored filter as a whole, so read it with `vision-scanners-get` and send it back in full with your changes. Common changes: toggle `enabled` to pause/resume scanning, adjust `sampling_rate` or `sampling_mode`, tweak the `scanner_config` prompt or tags, or refine the `query` to watch a different slice of sessions. `scanner_type` is locked after creation — to change it, delete and recreate. Editing any config field bumps `scanner_version`; existing observations keep a snapshot of the previous config for reproducibility. If you widen the `query`, raise `sampling_rate`, loosen `sampling_mode`, or move to a pricier `model`, re-check projected spend with `vision-scanners-estimate` and remaining budget with `vision-quota-get` first — the scanner resumes sweeping every 5 minutes at the new, broader scope.","category": "Replay vision","feature": "replay_vision","summary": "Update replay vision scanner",diff --git a/services/mcp/schema/tool-definitions-all.json b/services/mcp/schema/tool-definitions-all.jsonindex 1995b7533e68..fdcab781c8c5 100644--- a/services/mcp/schema/tool-definitions-all.json+++ b/services/mcp/schema/tool-definitions-all.json@@ -16587,7 +16587,7 @@"requires_ai_consent": true},"vision-scanners-estimate": {- "description": "Preview how many observations a proposed scanner would generate, and what they would cost, before saving it. Send a `query` (`RecordingsQuery` shape; `date_from`/`date_to` are ignored — the estimate uses a fixed 30-day lookback) plus the proposed `sampling_rate`, `sampling_mode` and `model`, and `scanner_id` when re-estimating an existing scanner so its own projection isn't double-counted. For an experiment scanner, pass `experiment` with the same `experiment_id` and `variants` you will save, so the estimate counts only exposed sessions. Returns `matched_sessions_in_window`, the `window_days` it was measured over, `estimated_observations_per_month`, `credits_per_observation`, `estimated_credits_per_month`, and `other_enabled_scanners_monthly_credits` (the org's other enabled scanners' projected spend). Enabled scanners sweep every 5 minutes, so a broad query can exhaust the budget quickly — run this before `vision-scanners-create` and compare the *credit* figures (never raw observation counts) against `remaining` from `vision-quota-get`. These projections are per month while `remaining` covers only the rest of the current period, so prorate them by the days left between `period_start` and `period_end` before comparing.",+ "description": "Preview how many observations a proposed scanner would generate, and what they would cost, before saving it. Send a `query` (`RecordingsQuery` shape; `date_from`/`date_to` are ignored, because the estimate counts a recent window, `window_days` in the response, and scales it to 30 days) plus the proposed `sampling_rate`, `sampling_mode` and `model`, and `scanner_id` when re-estimating an existing scanner so its own projection isn't double-counted. For an experiment scanner, pass `experiment` with the same `experiment_id` and `variants` you will save, so the estimate counts only exposed sessions. Returns `matched_sessions_in_window`, the `window_days` it was measured over, `estimated_observations_per_month`, `credits_per_observation`, `estimated_credits_per_month`, and `other_enabled_scanners_monthly_credits` (the org's other enabled scanners' projected spend). Enabled scanners sweep every 5 minutes, so a broad query can exhaust the budget quickly — run this before `vision-scanners-create` and compare the *credit* figures (never raw observation counts) against `remaining` from `vision-quota-get`. These projections are per month while `remaining` covers only the rest of the current period, so prorate them by the days left between `period_start` and `period_end` before comparing.","category": "Replay vision","feature": "replay_vision","summary": "Estimate replay vision scanner volume",@@ -16831,7 +16831,7 @@}},"vision-scanners-update": {- "description": "Partially update a replay vision scanner by ID. Send only the fields you want to change. Common changes: toggle `enabled` to pause/resume scanning, adjust `sampling_rate` or `sampling_mode`, tweak the `scanner_config` prompt or tags, or refine the `query` to watch a different slice of sessions. `scanner_type` is locked after creation — to change it, delete and recreate. Editing any config field bumps `scanner_version`; existing observations keep a snapshot of the previous config for reproducibility. If you widen the `query`, raise `sampling_rate`, loosen `sampling_mode`, or move to a pricier `model`, re-check projected spend with `vision-scanners-estimate` and remaining budget with `vision-quota-get` first — the scanner resumes sweeping every 5 minutes at the new, broader scope.",+ "description": "Partially update a replay vision scanner by ID. Send only the fields you want to change, except `query`: it replaces the stored filter as a whole, so read it with `vision-scanners-get` and send it back in full with your changes. Common changes: toggle `enabled` to pause/resume scanning, adjust `sampling_rate` or `sampling_mode`, tweak the `scanner_config` prompt or tags, or refine the `query` to watch a different slice of sessions. `scanner_type` is locked after creation — to change it, delete and recreate. Editing any config field bumps `scanner_version`; existing observations keep a snapshot of the previous config for reproducibility. If you widen the `query`, raise `sampling_rate`, loosen `sampling_mode`, or move to a pricier `model`, re-check projected spend with `vision-scanners-estimate` and remaining budget with `vision-quota-get` first — the scanner resumes sweeping every 5 minutes at the new, broader scope.","category": "Replay vision","feature": "replay_vision","summary": "Update replay vision scanner",diff --git a/services/mcp/schema/tool-inputs.json b/services/mcp/schema/tool-inputs.jsonindex 9909f186b798..7fda6a07eacc 100644--- a/services/mcp/schema/tool-inputs.json+++ b/services/mcp/schema/tool-inputs.json@@ -1836,6 +1836,381 @@},"description": "The complete desired set of shared (saved) metrics for the experiment — this REPLACES all existing saved-metric links, it does not append. To add or remove one, first read the experiment's current saved_metrics via experiment-get and resend the full set. Pass an empty array to detach all shared metrics."},+ "ScannerEstimateRecordingsFilterSchema": {+ "description": "Which recordings to count, as a `RecordingsQuery`. Omit every filter to count every recording. The date range is ignored, because the estimate counts a recent window (`window_days` in the response) and scales it to 30 days.",+ "type": "object",+ "properties": {+ "kind": {+ "default": "RecordingsQuery",+ "type": "string",+ "const": "RecordingsQuery"+ },+ "events": {+ "description": "Events a recording must contain, such as `{\"id\": \"$pageview\", \"type\": \"events\", \"order\": 0}`. Each can carry `properties` that filter that event.",+ "anyOf": [+ {+ "type": "array",+ "items": {+ "type": "object",+ "propertyNames": {+ "type": "string"+ },+ "additionalProperties": {}+ }+ },+ {+ "type": "null"+ }+ ]+ },+ "actions": {+ "description": "Saved actions a recording must contain, such as `{\"id\": 12, \"type\": \"actions\", \"order\": 0}`.",+ "anyOf": [+ {+ "type": "array",+ "items": {+ "type": "object",+ "propertyNames": {+ "type": "string"+ },+ "additionalProperties": {}+ }+ },+ {+ "type": "null"+ }+ ]+ },+ "properties": {+ "description": "Recording, session, person and cohort filters, such as `{\"type\": \"recording\", \"key\": \"visited_page\", \"value\": \"/checkout\", \"operator\": \"icontains\"}` or `{\"type\": \"cohort\", \"key\": \"id\", \"value\": 7, \"operator\": \"in\"}`.",+ "anyOf": [+ {+ "type": "array",+ "items": {+ "type": "object",+ "propertyNames": {+ "type": "string"+ },+ "additionalProperties": {}+ }+ },+ {+ "type": "null"+ }+ ]+ },+ "having_predicates": {+ "description": "Duration filters, such as `{\"type\": \"recording\", \"key\": \"active_seconds\", \"value\": 60, \"operator\": \"gt\"}`. These always apply with AND, whatever `operand` says.",+ "anyOf": [+ {+ "type": "array",+ "items": {+ "type": "object",+ "propertyNames": {+ "type": "string"+ },+ "additionalProperties": {}+ }+ },+ {+ "type": "null"+ }+ ]+ },+ "console_log_filters": {+ "description": "Console log filters, such as `{\"type\": \"log_entry\", \"key\": \"level\", \"value\": [\"error\"], \"operator\": \"exact\"}`.",+ "anyOf": [+ {+ "type": "array",+ "items": {+ "type": "object",+ "propertyNames": {+ "type": "string"+ },+ "additionalProperties": {}+ }+ },+ {+ "type": "null"+ }+ ]+ },+ "operand": {+ "description": "How `events`, `actions`, `properties` and `console_log_filters` combine. `AND`, the default when omitted, keeps recordings that match every condition. `OR` keeps recordings that match any one. An explicit `null` is read as `OR`. `having_predicates` and `filter_test_accounts` always apply with AND.",+ "anyOf": [+ {+ "type": "string",+ "enum": ["AND", "OR"]+ },+ {+ "type": "null"+ }+ ]+ },+ "filter_test_accounts": {+ "description": "Leave out recordings from internal and test accounts.",+ "anyOf": [+ {+ "type": "boolean"+ },+ {+ "type": "null"+ }+ ]+ }+ },+ "additionalProperties": {}+ },+ "ScannerRecordingsFilterSchema": {+ "description": "Which recordings the scanner watches, as a `RecordingsQuery`. Omit every filter to watch every new recording. The date range is ignored, because the schedule controls time.",+ "type": "object",+ "properties": {+ "kind": {+ "default": "RecordingsQuery",+ "type": "string",+ "const": "RecordingsQuery"+ },+ "events": {+ "description": "Events a recording must contain, such as `{\"id\": \"$pageview\", \"type\": \"events\", \"order\": 0}`. Each can carry `properties` that filter that event.",+ "anyOf": [+ {+ "type": "array",+ "items": {+ "type": "object",+ "propertyNames": {+ "type": "string"+ },+ "additionalProperties": {}+ }+ },+ {+ "type": "null"+ }+ ]+ },+ "actions": {+ "description": "Saved actions a recording must contain, such as `{\"id\": 12, \"type\": \"actions\", \"order\": 0}`.",+ "anyOf": [+ {+ "type": "array",+ "items": {+ "type": "object",+ "propertyNames": {+ "type": "string"+ },+ "additionalProperties": {}+ }+ },+ {+ "type": "null"+ }+ ]+ },+ "properties": {+ "description": "Recording, session, person and cohort filters, such as `{\"type\": \"recording\", \"key\": \"visited_page\", \"value\": \"/checkout\", \"operator\": \"icontains\"}` or `{\"type\": \"cohort\", \"key\": \"id\", \"value\": 7, \"operator\": \"in\"}`.",+ "anyOf": [+ {+ "type": "array",+ "items": {+ "type": "object",+ "propertyNames": {+ "type": "string"+ },+ "additionalProperties": {}+ }+ },+ {+ "type": "null"+ }+ ]+ },+ "having_predicates": {+ "description": "Duration filters, such as `{\"type\": \"recording\", \"key\": \"active_seconds\", \"value\": 60, \"operator\": \"gt\"}`. These always apply with AND, whatever `operand` says.",+ "anyOf": [+ {+ "type": "array",+ "items": {+ "type": "object",+ "propertyNames": {+ "type": "string"+ },+ "additionalProperties": {}+ }+ },+ {+ "type": "null"+ }+ ]+ },+ "console_log_filters": {+ "description": "Console log filters, such as `{\"type\": \"log_entry\", \"key\": \"level\", \"value\": [\"error\"], \"operator\": \"exact\"}`.",+ "anyOf": [+ {+ "type": "array",+ "items": {+ "type": "object",+ "propertyNames": {+ "type": "string"+ },+ "additionalProperties": {}+ }+ },+ {+ "type": "null"+ }+ ]+ },+ "operand": {+ "description": "How `events`, `actions`, `properties` and `console_log_filters` combine. `AND`, the default when omitted, keeps recordings that match every condition. `OR` keeps recordings that match any one. An explicit `null` is read as `OR`. `having_predicates` and `filter_test_accounts` always apply with AND.",+ "anyOf": [+ {+ "type": "string",+ "enum": ["AND", "OR"]+ },+ {+ "type": "null"+ }+ ]+ },+ "filter_test_accounts": {+ "description": "Leave out recordings from internal and test accounts.",+ "anyOf": [+ {+ "type": "boolean"+ },+ {+ "type": "null"+ }+ ]+ }+ },+ "additionalProperties": {}+ },+ "ScannerUpdateRecordingsFilterSchema": {+ "description": "The filter that replaces the scanner's current one, as a `RecordingsQuery`. It is not merged: any field you leave out is cleared, and `{}` watches every new recording. To change part of the filter, read the current one with `vision-scanners-get`, change it, and send it back in full. The date range is ignored, because the schedule controls time.",+ "type": "object",+ "properties": {+ "kind": {+ "default": "RecordingsQuery",+ "type": "string",+ "const": "RecordingsQuery"+ },+ "events": {+ "description": "Events a recording must contain, such as `{\"id\": \"$pageview\", \"type\": \"events\", \"order\": 0}`. Each can carry `properties` that filter that event.",+ "anyOf": [+ {+ "type": "array",+ "items": {+ "type": "object",+ "propertyNames": {+ "type": "string"+ },+ "additionalProperties": {}+ }+ },+ {+ "type": "null"+ }+ ]+ },+ "actions": {+ "description": "Saved actions a recording must contain, such as `{\"id\": 12, \"type\": \"actions\", \"order\": 0}`.",+ "anyOf": [+ {+ "type": "array",+ "items": {+ "type": "object",+ "propertyNames": {+ "type": "string"+ },+ "additionalProperties": {}+ }+ },+ {+ "type": "null"+ }+ ]+ },+ "properties": {+ "description": "Recording, session, person and cohort filters, such as `{\"type\": \"recording\", \"key\": \"visited_page\", \"value\": \"/checkout\", \"operator\": \"icontains\"}` or `{\"type\": \"cohort\", \"key\": \"id\", \"value\": 7, \"operator\": \"in\"}`.",+ "anyOf": [+ {+ "type": "array",+ "items": {+ "type": "object",+ "propertyNames": {+ "type": "string"+ },+ "additionalProperties": {}+ }+ },+ {+ "type": "null"+ }+ ]+ },+ "having_predicates": {+ "description": "Duration filters, such as `{\"type\": \"recording\", \"key\": \"active_seconds\", \"value\": 60, \"operator\": \"gt\"}`. These always apply with AND, whatever `operand` says.",+ "anyOf": [+ {+ "type": "array",+ "items": {+ "type": "object",+ "propertyNames": {+ "type": "string"+ },+ "additionalProperties": {}+ }+ },+ {+ "type": "null"+ }+ ]+ },+ "console_log_filters": {+ "description": "Console log filters, such as `{\"type\": \"log_entry\", \"key\": \"level\", \"value\": [\"error\"], \"operator\": \"exact\"}`.",+ "anyOf": [+ {+ "type": "array",+ "items": {+ "type": "object",+ "propertyNames": {+ "type": "string"+ },+ "additionalProperties": {}+ }+ },+ {+ "type": "null"+ }+ ]+ },+ "operand": {+ "description": "How `events`, `actions`, `properties` and `console_log_filters` combine. `AND`, the default when omitted, keeps recordings that match every condition. `OR` keeps recordings that match any one. An explicit `null` is read as `OR`. `having_predicates` and `filter_test_accounts` always apply with AND.",+ "anyOf": [+ {+ "type": "string",+ "enum": ["AND", "OR"]+ },+ {+ "type": "null"+ }+ ]+ },+ "filter_test_accounts": {+ "description": "Leave out recordings from internal and test accounts.",+ "anyOf": [+ {+ "type": "boolean"+ },+ {+ "type": "null"+ }+ ]+ }+ },+ "additionalProperties": {}+ },"ScoreDefinitionConfigSchema": {"anyOf": [{diff --git a/services/mcp/src/generated/replay_vision/api.ts b/services/mcp/src/generated/replay_vision/api.tsindex 664dfebaa07f..cdda1ef3b822 100644--- a/services/mcp/src/generated/replay_vision/api.ts+++ b/services/mcp/src/generated/replay_vision/api.ts@@ -1073,12 +1073,6 @@ export const VisionScannersCreateBody = () => zod.describe('Type-specific configuration. All scanner types require `prompt`; monitors add optional `allow_inconclusive`, classifiers add `tags`, scorers add `scale`, summarizers add optional `length`.'),- query: zod- .unknown()- .optional()- .describe(- 'Persisted `RecordingsQuery` shape used to pick candidate sessions. `date_from`\/`date_to` are stripped on save — the schedule controls time, not the user.'- ),sampling_rate: zod.number().min(visionScannersCreateBodySamplingRateMin)@@ -1258,12 +1252,6 @@ export const VisionScannersPartialUpdateBody = () => zod.describe('Type-specific configuration. All scanner types require `prompt`; monitors add optional `allow_inconclusive`, classifiers add `tags`, scorers add `scale`, summarizers add optional `length`.'),- query: zod- .unknown()- .optional()- .describe(- 'Persisted `RecordingsQuery` shape used to pick candidate sessions. `date_from`\/`date_to` are stripped on save — the schedule controls time, not the user.'- ),sampling_rate: zod.number().min(visionScannersPartialUpdateBodySamplingRateMin)@@ -2433,12 +2421,6 @@ export const visionScannersEstimateCreateBodyExperimentOneVariantsItemMax = 400export const VisionScannersEstimateCreateBody = () => zod.object({- query: zod- .unknown()- .optional()- .describe(- 'Proposed `RecordingsQuery` for the candidate filter. `date_from`\/`date_to` are ignored — the estimate scans a recent window (`window_days` in the response) and scales it to 30 days. Omit to estimate against all recordings.'- ),sampling_rate: zod.number().min(visionScannersEstimateCreateBodySamplingRateMin)diff --git a/services/mcp/src/schema/tool-inputs.ts b/services/mcp/src/schema/tool-inputs.tsindex eb664342ea7e..827b35bb4ec1 100644--- a/services/mcp/src/schema/tool-inputs.ts+++ b/services/mcp/src/schema/tool-inputs.ts@@ -1497,3 +1497,48 @@ export const FlagRulesV2UpdateInputSchema = z.preprocess(.describe('The complete new configuration. It replaces the stored `filters`.'),}))++// A scanner's recording filter, as the replay vision tools take it. The API validates the full+// `RecordingsQuery`; inlining that schema in every scanner tool costs agents hundreds of kilobytes each,+// so this names the parts a scanner filters on and lets each item through for the API to check.+const RecordingFilterItemsSchema = z.array(z.record(z.string(), z.unknown()))++const ScannerRecordingsFilterFields = z+ .object({+ kind: z.literal('RecordingsQuery').default('RecordingsQuery'),+ events: RecordingFilterItemsSchema.nullish().describe(+ 'Events a recording must contain, such as `{"id": "$pageview", "type": "events", "order": 0}`. Each can carry `properties` that filter that event.'+ ),+ actions: RecordingFilterItemsSchema.nullish().describe(+ 'Saved actions a recording must contain, such as `{"id": 12, "type": "actions", "order": 0}`.'+ ),+ properties: RecordingFilterItemsSchema.nullish().describe(+ 'Recording, session, person and cohort filters, such as `{"type": "recording", "key": "visited_page", "value": "/checkout", "operator": "icontains"}` or `{"type": "cohort", "key": "id", "value": 7, "operator": "in"}`.'+ ),+ having_predicates: RecordingFilterItemsSchema.nullish().describe(+ 'Duration filters, such as `{"type": "recording", "key": "active_seconds", "value": 60, "operator": "gt"}`. These always apply with AND, whatever `operand` says.'+ ),+ console_log_filters: RecordingFilterItemsSchema.nullish().describe(+ 'Console log filters, such as `{"type": "log_entry", "key": "level", "value": ["error"], "operator": "exact"}`.'+ ),+ operand: z+ .enum(['AND', 'OR'])+ .nullish()+ .describe(+ 'How `events`, `actions`, `properties` and `console_log_filters` combine. `AND`, the default when omitted, keeps recordings that match every condition. `OR` keeps recordings that match any one. An explicit `null` is read as `OR`. `having_predicates` and `filter_test_accounts` always apply with AND.'+ ),+ filter_test_accounts: z.boolean().nullish().describe('Leave out recordings from internal and test accounts.'),+ })+ .passthrough()++export const ScannerRecordingsFilterSchema = ScannerRecordingsFilterFields.optional().describe(+ 'Which recordings the scanner watches, as a `RecordingsQuery`. Omit every filter to watch every new recording. The date range is ignored, because the schedule controls time.'+)++export const ScannerUpdateRecordingsFilterSchema = ScannerRecordingsFilterFields.optional().describe(+ "The filter that replaces the scanner's current one, as a `RecordingsQuery`. It is not merged: any field you leave out is cleared, and `{}` watches every new recording. To change part of the filter, read the current one with `vision-scanners-get`, change it, and send it back in full. The date range is ignored, because the schedule controls time."+)++export const ScannerEstimateRecordingsFilterSchema = ScannerRecordingsFilterFields.optional().describe(+ 'Which recordings to count, as a `RecordingsQuery`. Omit every filter to count every recording. The date range is ignored, because the estimate counts a recent window (`window_days` in the response) and scales it to 30 days.'+)diff --git a/services/mcp/src/tools/generated/replay_vision.ts b/services/mcp/src/tools/generated/replay_vision.tsindex 0fb66150aa17..dd22cce973a4 100644--- a/services/mcp/src/tools/generated/replay_vision.ts+++ b/services/mcp/src/tools/generated/replay_vision.ts@@ -4,6 +4,11 @@ import { z } from 'zod'import type { Schemas } from '@/api/generated'import * as orvalSchemas from '@/generated/replay_vision/api'import { withUiApp } from '@/resources/ui-apps'+import {+ ScannerEstimateRecordingsFilterSchema,+ ScannerRecordingsFilterSchema,+ ScannerUpdateRecordingsFilterSchema,+} from '@/schema/tool-inputs'import { castBooleanToString } from '@/tools/cast-helpers'import {withPostHogUrl,@@ -959,7 +964,7 @@ const visionScannersCounts = (): ToolBase<const VisionScannersCreateSchema = () => {const VisionScannersCreateBody = orvalSchemas.VisionScannersCreateBody()- return VisionScannersCreateBody+ return VisionScannersCreateBody.extend({ query: ScannerRecordingsFilterSchema })}const visionScannersCreate = (): ToolBase<@@ -995,9 +1000,6 @@ const visionScannersCreate = (): ToolBase<if (params.scanner_config !== undefined) {body['scanner_config'] = params.scanner_config}- if (params.query !== undefined) {- body['query'] = params.query- }if (params.sampling_rate !== undefined) {body['sampling_rate'] = params.sampling_rate}@@ -1025,6 +1027,9 @@ const visionScannersCreate = (): ToolBase<if (params.experiment_targeting !== undefined) {body['experiment_targeting'] = params.experiment_targeting}+ if (params.query !== undefined) {+ body['query'] = params.query+ }const result = await context.api.request<Schemas.ReplayScanner>({method: 'POST',path: `/api/projects/${encodeURIComponent(String(projectId))}/vision/scanners/`,@@ -1137,7 +1142,7 @@ const visionScannersDuplicate = (): ToolBase<const VisionScannersEstimateSchema = () => {const VisionScannersEstimateCreateBody = orvalSchemas.VisionScannersEstimateCreateBody()- return VisionScannersEstimateCreateBody+ return VisionScannersEstimateCreateBody.extend({ query: ScannerEstimateRecordingsFilterSchema })}const visionScannersEstimate = (): ToolBase<@@ -1149,9 +1154,6 @@ const visionScannersEstimate = (): ToolBase<handler: async (context: Context, params: z.infer<ReturnType<typeof VisionScannersEstimateSchema>>) => {const projectId = await context.stateManager.getProjectId()const body: Record<string, unknown> = {}- if (params.query !== undefined) {- body['query'] = params.query- }if (params.sampling_rate !== undefined) {body['sampling_rate'] = params.sampling_rate}@@ -1170,6 +1172,9 @@ const visionScannersEstimate = (): ToolBase<if (params.experiment !== undefined) {body['experiment'] = params.experiment}+ if (params.query !== undefined) {+ body['query'] = params.query+ }const result = await context.api.request<Schemas.EstimateResponse>({method: 'POST',path: `/api/projects/${encodeURIComponent(String(projectId))}/vision/scanners/estimate/`,@@ -1700,7 +1705,9 @@ const visionScannersSelfDrivingStats = (): ToolBase<const VisionScannersUpdateSchema = () => {const VisionScannersPartialUpdateBody = orvalSchemas.VisionScannersPartialUpdateBody()const VisionScannersPartialUpdateParams = orvalSchemas.VisionScannersPartialUpdateParams()- return VisionScannersPartialUpdateParams.omit({ project_id: true }).extend(VisionScannersPartialUpdateBody.shape)+ return VisionScannersPartialUpdateParams.omit({ project_id: true })+ .extend(VisionScannersPartialUpdateBody.shape)+ .extend({ query: ScannerUpdateRecordingsFilterSchema.optional() })}const visionScannersUpdate = (): ToolBase<ReturnType<typeof VisionScannersUpdateSchema>, Schemas.ReplayScanner> => ({@@ -1733,9 +1740,6 @@ const visionScannersUpdate = (): ToolBase<ReturnType<typeof VisionScannersUpdateif (params.scanner_config !== undefined) {body['scanner_config'] = params.scanner_config}- if (params.query !== undefined) {- body['query'] = params.query- }if (params.sampling_rate !== undefined) {body['sampling_rate'] = params.sampling_rate}@@ -1763,6 +1767,9 @@ const visionScannersUpdate = (): ToolBase<ReturnType<typeof VisionScannersUpdateif (params.experiment_targeting !== undefined) {body['experiment_targeting'] = params.experiment_targeting}+ if (params.query !== undefined) {+ body['query'] = params.query+ }const result = await context.api.request<Schemas.ReplayScanner>({method: 'PATCH',path: `/api/projects/${encodeURIComponent(String(projectId))}/vision/scanners/${encodeURIComponent(String(params.id))}/`,diff --git a/services/mcp/tests/unit/__snapshots__/tool-schemas/vision-scanners-create.json b/services/mcp/tests/unit/__snapshots__/tool-schemas/vision-scanners-create.jsonindex c3ce5ae708ef..2490eeeb36df 100644--- a/services/mcp/tests/unit/__snapshots__/tool-schemas/vision-scanners-create.json+++ b/services/mcp/tests/unit/__snapshots__/tool-schemas/vision-scanners-create.json@@ -1,6 +1,5 @@{"$schema": "https://json-schema.org/draft/2020-12/schema",- "description": "A Replay Vision scanner: its type, targeting query, and AI configuration.","properties": {"creation_method": {"anyOf": [@@ -108,7 +107,129 @@"type": "string"},"query": {- "description": "Persisted `RecordingsQuery` shape used to pick candidate sessions. `date_from`/`date_to` are stripped on save — the schedule controls time, not the user."+ "additionalProperties": {},+ "description": "Which recordings the scanner watches, as a `RecordingsQuery`. Omit every filter to watch every new recording. The date range is ignored, because the schedule controls time.",+ "properties": {+ "actions": {+ "anyOf": [+ {+ "items": {+ "additionalProperties": {},+ "propertyNames": {+ "type": "string"+ },+ "type": "object"+ },+ "type": "array"+ },+ {+ "type": "null"+ }+ ],+ "description": "Saved actions a recording must contain, such as `{\"id\": 12, \"type\": \"actions\", \"order\": 0}`."+ },+ "console_log_filters": {+ "anyOf": [+ {+ "items": {+ "additionalProperties": {},+ "propertyNames": {+ "type": "string"+ },+ "type": "object"+ },+ "type": "array"+ },+ {+ "type": "null"+ }+ ],+ "description": "Console log filters, such as `{\"type\": \"log_entry\", \"key\": \"level\", \"value\": [\"error\"], \"operator\": \"exact\"}`."+ },+ "events": {+ "anyOf": [+ {+ "items": {+ "additionalProperties": {},+ "propertyNames": {+ "type": "string"+ },+ "type": "object"+ },+ "type": "array"+ },+ {+ "type": "null"+ }+ ],+ "description": "Events a recording must contain, such as `{\"id\": \"$pageview\", \"type\": \"events\", \"order\": 0}`. Each can carry `properties` that filter that event."+ },+ "filter_test_accounts": {+ "anyOf": [+ {+ "type": "boolean"+ },+ {+ "type": "null"+ }+ ],+ "description": "Leave out recordings from internal and test accounts."+ },+ "having_predicates": {+ "anyOf": [+ {+ "items": {+ "additionalProperties": {},+ "propertyNames": {+ "type": "string"+ },+ "type": "object"+ },+ "type": "array"+ },+ {+ "type": "null"+ }+ ],+ "description": "Duration filters, such as `{\"type\": \"recording\", \"key\": \"active_seconds\", \"value\": 60, \"operator\": \"gt\"}`. These always apply with AND, whatever `operand` says."+ },+ "kind": {+ "const": "RecordingsQuery",+ "default": "RecordingsQuery",+ "type": "string"+ },+ "operand": {+ "anyOf": [+ {+ "enum": ["AND", "OR"],+ "type": "string"+ },+ {+ "type": "null"+ }+ ],+ "description": "How `events`, `actions`, `properties` and `console_log_filters` combine. `AND`, the default when omitted, keeps recordings that match every condition. `OR` keeps recordings that match any one. An explicit `null` is read as `OR`. `having_predicates` and `filter_test_accounts` always apply with AND."+ },+ "properties": {+ "anyOf": [+ {+ "items": {+ "additionalProperties": {},+ "propertyNames": {+ "type": "string"+ },+ "type": "object"+ },+ "type": "array"+ },+ {+ "type": "null"+ }+ ],+ "description": "Recording, session, person and cohort filters, such as `{\"type\": \"recording\", \"key\": \"visited_page\", \"value\": \"/checkout\", \"operator\": \"icontains\"}` or `{\"type\": \"cohort\", \"key\": \"id\", \"value\": 7, \"operator\": \"in\"}`."+ }+ },+ "type": "object"},"sampling_mode": {"description": "Quality pre-filter applied before random sampling. focused = top sessions only, balanced = drops the lowest-quality, comprehensive = no filter (default).\n\n* `focused` - Focused\n* `balanced` - Balanced\n* `comprehensive` - Comprehensive",diff --git a/services/mcp/tests/unit/__snapshots__/tool-schemas/vision-scanners-estimate.json b/services/mcp/tests/unit/__snapshots__/tool-schemas/vision-scanners-estimate.jsonindex ba578daf7927..0c16cd539b93 100644--- a/services/mcp/tests/unit/__snapshots__/tool-schemas/vision-scanners-estimate.json+++ b/services/mcp/tests/unit/__snapshots__/tool-schemas/vision-scanners-estimate.json@@ -1,6 +1,5 @@{"$schema": "https://json-schema.org/draft/2020-12/schema",- "description": "Body of POST /vision/scanners/estimate/ — a proposed, unsaved scanner config.","properties": {"experiment": {"anyOf": [@@ -79,7 +78,129 @@"type": "string"},"query": {- "description": "Proposed `RecordingsQuery` for the candidate filter. `date_from`/`date_to` are ignored — the estimate scans a recent window (`window_days` in the response) and scales it to 30 days. Omit to estimate against all recordings."+ "additionalProperties": {},+ "description": "Which recordings to count, as a `RecordingsQuery`. Omit every filter to count every recording. The date range is ignored, because the estimate counts a recent window (`window_days` in the response) and scales it to 30 days.",+ "properties": {+ "actions": {+ "anyOf": [+ {+ "items": {+ "additionalProperties": {},+ "propertyNames": {+ "type": "string"+ },+ "type": "object"+ },+ "type": "array"+ },+ {+ "type": "null"+ }+ ],+ "description": "Saved actions a recording must contain, such as `{\"id\": 12, \"type\": \"actions\", \"order\": 0}`."+ },+ "console_log_filters": {+ "anyOf": [+ {+ "items": {+ "additionalProperties": {},+ "propertyNames": {+ "type": "string"+ },+ "type": "object"+ },+ "type": "array"+ },+ {+ "type": "null"+ }+ ],+ "description": "Console log filters, such as `{\"type\": \"log_entry\", \"key\": \"level\", \"value\": [\"error\"], \"operator\": \"exact\"}`."+ },+ "events": {+ "anyOf": [+ {+ "items": {+ "additionalProperties": {},+ "propertyNames": {+ "type": "string"+ },+ "type": "object"+ },+ "type": "array"+ },+ {+ "type": "null"+ }+ ],+ "description": "Events a recording must contain, such as `{\"id\": \"$pageview\", \"type\": \"events\", \"order\": 0}`. Each can carry `properties` that filter that event."+ },+ "filter_test_accounts": {+ "anyOf": [+ {+ "type": "boolean"+ },+ {+ "type": "null"+ }+ ],+ "description": "Leave out recordings from internal and test accounts."+ },+ "having_predicates": {+ "anyOf": [+ {+ "items": {+ "additionalProperties": {},+ "propertyNames": {+ "type": "string"+ },+ "type": "object"+ },+ "type": "array"+ },+ {+ "type": "null"+ }+ ],+ "description": "Duration filters, such as `{\"type\": \"recording\", \"key\": \"active_seconds\", \"value\": 60, \"operator\": \"gt\"}`. These always apply with AND, whatever `operand` says."+ },+ "kind": {+ "const": "RecordingsQuery",+ "default": "RecordingsQuery",+ "type": "string"+ },+ "operand": {+ "anyOf": [+ {+ "enum": ["AND", "OR"],+ "type": "string"+ },+ {+ "type": "null"+ }+ ],+ "description": "How `events`, `actions`, `properties` and `console_log_filters` combine. `AND`, the default when omitted, keeps recordings that match every condition. `OR` keeps recordings that match any one. An explicit `null` is read as `OR`. `having_predicates` and `filter_test_accounts` always apply with AND."+ },+ "properties": {+ "anyOf": [+ {+ "items": {+ "additionalProperties": {},+ "propertyNames": {+ "type": "string"+ },+ "type": "object"+ },+ "type": "array"+ },+ {+ "type": "null"+ }+ ],+ "description": "Recording, session, person and cohort filters, such as `{\"type\": \"recording\", \"key\": \"visited_page\", \"value\": \"/checkout\", \"operator\": \"icontains\"}` or `{\"type\": \"cohort\", \"key\": \"id\", \"value\": 7, \"operator\": \"in\"}`."+ }+ },+ "type": "object"},"sampling_mode": {"default": "comprehensive",diff --git a/services/mcp/tests/unit/__snapshots__/tool-schemas/vision-scanners-update.json b/services/mcp/tests/unit/__snapshots__/tool-schemas/vision-scanners-update.jsonindex a2b5ffe0ff73..05e8e7f85b23 100644--- a/services/mcp/tests/unit/__snapshots__/tool-schemas/vision-scanners-update.json+++ b/services/mcp/tests/unit/__snapshots__/tool-schemas/vision-scanners-update.json@@ -111,7 +111,129 @@"type": "string"},"query": {- "description": "Persisted `RecordingsQuery` shape used to pick candidate sessions. `date_from`/`date_to` are stripped on save — the schedule controls time, not the user."+ "additionalProperties": {},+ "description": "The filter that replaces the scanner's current one, as a `RecordingsQuery`. It is not merged: any field you leave out is cleared, and `{}` watches every new recording. To change part of the filter, read the current one with `vision-scanners-get`, change it, and send it back in full. The date range is ignored, because the schedule controls time.",+ "properties": {+ "actions": {+ "anyOf": [+ {+ "items": {+ "additionalProperties": {},+ "propertyNames": {+ "type": "string"+ },+ "type": "object"+ },+ "type": "array"+ },+ {+ "type": "null"+ }+ ],+ "description": "Saved actions a recording must contain, such as `{\"id\": 12, \"type\": \"actions\", \"order\": 0}`."+ },+ "console_log_filters": {+ "anyOf": [+ {+ "items": {+ "additionalProperties": {},+ "propertyNames": {+ "type": "string"+ },+ "type": "object"+ },+ "type": "array"+ },+ {+ "type": "null"+ }+ ],+ "description": "Console log filters, such as `{\"type\": \"log_entry\", \"key\": \"level\", \"value\": [\"error\"], \"operator\": \"exact\"}`."+ },+ "events": {+ "anyOf": [+ {+ "items": {+ "additionalProperties": {},+ "propertyNames": {+ "type": "string"+ },+ "type": "object"+ },+ "type": "array"+ },+ {+ "type": "null"+ }+ ],+ "description": "Events a recording must contain, such as `{\"id\": \"$pageview\", \"type\": \"events\", \"order\": 0}`. Each can carry `properties` that filter that event."+ },+ "filter_test_accounts": {+ "anyOf": [+ {+ "type": "boolean"+ },+ {+ "type": "null"+ }+ ],+ "description": "Leave out recordings from internal and test accounts."+ },+ "having_predicates": {+ "anyOf": [+ {+ "items": {+ "additionalProperties": {},+ "propertyNames": {+ "type": "string"+ },+ "type": "object"+ },+ "type": "array"+ },+ {+ "type": "null"+ }+ ],+ "description": "Duration filters, such as `{\"type\": \"recording\", \"key\": \"active_seconds\", \"value\": 60, \"operator\": \"gt\"}`. These always apply with AND, whatever `operand` says."+ },+ "kind": {+ "const": "RecordingsQuery",+ "default": "RecordingsQuery",+ "type": "string"+ },+ "operand": {+ "anyOf": [+ {+ "enum": ["AND", "OR"],+ "type": "string"+ },+ {+ "type": "null"+ }+ ],+ "description": "How `events`, `actions`, `properties` and `console_log_filters` combine. `AND`, the default when omitted, keeps recordings that match every condition. `OR` keeps recordings that match any one. An explicit `null` is read as `OR`. `having_predicates` and `filter_test_accounts` always apply with AND."+ },+ "properties": {+ "anyOf": [+ {+ "items": {+ "additionalProperties": {},+ "propertyNames": {+ "type": "string"+ },+ "type": "object"+ },+ "type": "array"+ },+ {+ "type": "null"+ }+ ],+ "description": "Recording, session, person and cohort filters, such as `{\"type\": \"recording\", \"key\": \"visited_page\", \"value\": \"/checkout\", \"operator\": \"icontains\"}` or `{\"type\": \"cohort\", \"key\": \"id\", \"value\": 7, \"operator\": \"in\"}`."+ }+ },+ "type": "object"},"sampling_mode": {"description": "Quality pre-filter applied before random sampling. focused = top sessions only, balanced = drops the lowest-quality, comprehensive = no filter (default).\n\n* `focused` - Focused\n* `balanced` - Balanced\n* `comprehensive` - Comprehensive",