Skip to main content
[ ARKIV ]HUB
Tools
Tools · proposed · ObservabilityNot yet

See every write, error, and what it cost — privately.

A thin wrapper around the SDK that captures metadata about your usage — method, latency, error, measured cost — and never the payload. It runs continuously inside the app (category a), so a skill or the MCP can't do it. You choose the sink: off-chain by default, or a dogfooding dashboard stored as metadata-only entities on Arkiv.

arkiv-telemetrytelemetryruntimecostprivacy
Status
Not yet built
Type
Tool
Feasibility
Medium
THE IDEA

SDK middleware that records what your app does with Arkiv — methods, errors, latency, measured GLM cost — with payloads and secrets stripped, so you get observability without leaking data onto a public chain.

The problem today

  • Developers can't see their own usage: which methods they call, what's failing, or what a write actually costs.
  • Arkiv's pricing calculator is missing (blocked by an incomplete pricing model), so cost is a black box.
  • Naively "log everything to Arkiv" would leak prompts, PII, and secrets onto a public chain — telemetry must be metadata-only and sanitized.

What it would do

  • Wraps the client: logs method, attribute KEYS (not values), payload SIZE (not content), latency, error class, and measured cost per call.
  • Sink is configurable: off-chain (file/console/OTel) by default; an opt-in dogfooding mode stores sanitized, metadata-only telemetry on Arkiv.
  • Cost estimator: given a workload (entity count, payload size, TTL, batch size), estimates GLM from testnet-measured rates — measured cost, not official pricing.
WHY IT’S A BETTER EXPERIENCE
  • "Trust through visibility" — devs trust a system when they can see spend, failures, and latency (Supabase Logs / Convex insights).
  • Privacy-first by design: it physically can't ship payloads or secrets, which a public chain demands.
  • Gives DevRel the structured friction report that justifies every event budget.
HOW IT COMPARES
Supabase Logs / Metricsobservability is why teams feel safe in prod. Arkiv has none yet.
Helicone (AI proxy)change a base URL, get telemetry — the low-friction model, with sanitization built in.
FEASIBILITY — AND WHY THIS ISN’T HAND-WAVING
Medium to build

Logging is easy; the careful part is sanitization (never payloads/secrets) and honest "measured, not pricing" cost estimates.

Grounded in

  • Arkiv DevRel mandate is explicitly telemetry-first; "Dogfood What We Build" is a company value.
  • Pricing calculator is a known missing tool (handoff §13.2) — this fills the measurable part now.
  • Security review flagged the leak risk of logging to a public chain → metadata-only is mandatory.
PAIRS WITH
arkiv-telemetryThis is a proposal— not built yet. It’s here so the case can be read and the idea ranked. Want it? Open an issue on the Arkiv GitHub.
© 2026 Golem Factory · Braga testnet · Proposed tool — status: not yet.