Skip to content

: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.

3.6 KB
Raw Blame
# Keep in sync with frontend `BREAKPOINT_COLUMN_COUNTS.sm`.
DASHBOARD_GRID_COLUMN_COUNT = 12

# Hard cap on rows returned per widget query (run_widgets throttles + UI "25+" footer).
MAX_WIDGET_RESULT_LIMIT = 25

# Default when config omits limit — catalog/OpenAPI default, below the hard cap.
DEFAULT_WIDGET_LIST_LIMIT = 10

# Activity events list allows a larger page than other list widgets (lightweight rows).
ACTIVITY_EVENTS_MAX_LIMIT = 50
ACTIVITY_EVENTS_DEFAULT_LIMIT = 25
ACTIVITY_EVENTS_MAX_PROPERTY_FILTERS = 20
ACTIVITY_EVENTS_MAX_PROPERTY_FILTER_VALUES = 100
ACTIVITY_EVENTS_MAX_PROPERTY_KEY_LENGTH = 400
ACTIVITY_EVENTS_MAX_PROPERTY_LABEL_LENGTH = 400
ACTIVITY_EVENTS_MAX_PROPERTY_VALUE_LENGTH = 4000

# Logs list allows the largest page — log rows are cheap and users scan many at once.
LOGS_LIST_MAX_LIMIT = 100
LOGS_LIST_DEFAULT_LIMIT = 50

# Metrics table: one metrics query clause per column, and one row per value of the group-by label.
# The metrics query keeps at most 100 series per clause, so the row limit cannot go higher.
# Six columns fit rate, errors and three percentiles side by side.
METRICS_TABLE_MAX_COLUMNS = 6
METRICS_TABLE_MAX_FILTERS = 5
METRICS_TABLE_MAX_LIMIT = 100
METRICS_TABLE_DEFAULT_LIMIT = 25
# Points per sparkline, and also the bucket count of the query. A query returns one row per bucket and label
# value, and the metrics query fails above its row budget, so fewer buckets leave room for more label values.
METRICS_TABLE_SPARKLINE_POINTS = 30
# A label with many values can go past the metrics row budget at METRICS_TABLE_SPARKLINE_POINTS buckets.
# The widget then retries once with this many buckets: shorter sparklines, the same row values.
METRICS_TABLE_RETRY_BUCKETS = 6
# Metrics queries span at most 31 days, so the longer widget presets are not allowed.
METRICS_TABLE_UNSUPPORTED_DATE_FROM_VALUES = frozenset({"-90d"})

# Cap widgets per batch create / run-widgets request.
MAX_WIDGETS_BATCH_SIZE = 10

MAX_TEXT_TILE_AGENT_CONTEXT_LENGTH = 10_000
MAX_GROUPS = 100
MAX_GROUP_KEY_LENGTH = 100
MAX_GROUP_TITLE_LENGTH = 200

# Upper bound on pinned recordings pulled from a collection to scope a session replay widget. Bounds the
# session_ids IN clause sent to ClickHouse and the Python list we materialize, matching the saved-filter
# cap (MAX_SAVED_FILTER_SESSION_IDS_PER_PLAYLIST). The widget itself only ever shows MAX_WIDGET_RESULT_LIMIT.
MAX_COLLECTION_SESSION_IDS = 1000

# Preset relative ranges only — widgets don't accept arbitrary HogQL date strings.
# Note: `M` is minutes, `m` would be months (see posthog.utils relative-date parsing).
WIDGET_DATE_FROM_VALUES_ORDERED: tuple[str, ...] = (
    "-1M",
    "-30M",
    "-1h",
    "-3h",
    "-24h",
    "-7d",
    "-14d",
    "-30d",
    "-90d",
)

WIDGET_DATE_FROM_LABELS: dict[str, str] = {
    "-1M": "Last minute",
    "-30M": "Last 30 minutes",
    "-1h": "Last hour",
    "-3h": "Last 3 hours",
    "-24h": "Last 24 hours",
    "-7d": "Last 7 days",
    "-14d": "Last 14 days",
    "-30d": "Last 30 days",
    "-90d": "Last 90 days",
}

# `optimized` run_insights output is read by agents with a limited context window, so each tile's
# formatted table is held to this many characters unless the caller asks for more.
RUN_INSIGHTS_DEFAULT_MAX_RESULT_CHARS = 2000

# Ceiling on the whole `optimized` run_insights response. Tiles past it are not run.
RUN_INSIGHTS_MAX_TOTAL_CHARS = 30000

# Below this, a tile can only carry a sliced header, so it is reported as not run instead.
RUN_INSIGHTS_MIN_TILE_CHARS = 200

# The marker list grows with the dashboard, so it gets a ceiling of its own.
RUN_INSIGHTS_MAX_UNRUN_TILES = 25