Strix (STRIX) — telemetry
flowchart LR
IN["IN: scan lifecycle events\n(start, tool calls, completion)"] --> ACT["ACT: emit anonymous\nusage events"]
ACT --> VAL["VAL: STRIX_TELEMETRY\non/off check"]
VAL --> OUT["OUT: events sent to\nPostHog / Scarf"]
What this is
strix/telemetry sends anonymous, aggregate usage signals (not target data, not
findings, not code) back to the maintainers — the kind of "is this feature used"
counting most open-source CLI tools do, gated by a settings flag.
What flows through it
flowchart TD
Logging["logging.py\nsets up structured logs\n(strix_runs/<run>/logs)"] --> Common["_common.py\nshared event shaping"]
Common --> PostHog["posthog.py"]
Common --> Scarf["scarf.py"]
Technical details
flowchart LR
IN["IN: scan lifecycle events\n(start, tool calls, completion)"] --> ACT["ACT: emit anonymous\nusage events"]
ACT --> VAL["VAL: STRIX_TELEMETRY\non/off check"]
VAL --> OUT["OUT: events sent to\nPostHog / Scarf"]
Package: strix/telemetry/
Depends on: strix.config (TelemetrySettings).
Depended on by: strix.interface (main entry wires up scan-lifecycle telemetry
calls), strix.report (cost-callback registration touches usage logging).
Files at a glance
| File | Role |
|---|---|
logging.py |
setup_scan_logging(), set_scan_id() — structured per-run logging, distinct from anonymous usage telemetry. |
posthog.py |
PostHog event emission for anonymous product-usage analytics. |
scarf.py |
Scarf-based install/usage pinging (common for CLI tools measuring adoption). |
_common.py |
Shared event-shaping helpers used by both backends. |
Files & functions, explained
Entirely deterministic — this module never calls a model; it only decides whether
(via STRIX_TELEMETRY) and what shape of event to send about the fact that a scan
ran, never its content.
Related: STRIX — Architecture (HLD), interface.md.