Claude API Telemetry on ClickHouse: A Production Guide to Cost, Latency, and Error Analytics
Stream per-request Claude API telemetry into ClickHouse, build sub-second dashboards with materialized views, and detect cost spikes, retry loops, and silent failures with practical SQL recipes.
Building a Production Claude API Pipeline on Cloudflare Queues: Fault Tolerance, Backpressure, and Cost Control
A practical, code-first walkthrough for routing Claude API calls through Cloudflare Queues — covering producer/consumer code, retry-vs-DLQ branching, priority lanes, and token budgeting for production workloads.
Observability for Claude Code with OpenTelemetry — A Production-Grade Tracing Guide for Agentic Workflows
Trace Claude Code agent runs end to end with OpenTelemetry. Hook integration, per-tool spans, MCP propagation, cost attribution, and sampling patterns that survive thousands of runs per day.
Production Semantic Cache for Claude API — Similarity Thresholds, Pollution Defense, and What to Track
A production playbook for adding a semantic cache in front of Claude API — threshold tuning, multi-tenant isolation, pollution prevention, fallbacks, how it stacks with Anthropic's prompt caching, and the metrics that actually prove it works.
Debugging Claude Code Hooks in Production — Where to Start When Logs Are Missing
Your Claude Code hooks work locally but go silent in production. Here is the three-layer observability pattern I use on real projects, plus a 60-second minimal reproduction recipe for isolating failures.
Shadow Mode with Claude Agent SDK — Measuring Agent Accuracy on Live Traffic Without Touching Users
You want to ship an AI agent to production, but you can't measure its real accuracy without exposing real users. Shadow mode solves that paradox. This guide shows how to run a Claude Agent SDK agent alongside your existing workflow, log the deltas, and promote it step by step.
When Prompt Caching Was On but the Bill Didn't Move — Field Notes on Instrumenting cache_read to Close the Hit-Rate Gap
You added cache_control but your Claude API bill barely changed. Before guessing at fixes, record cache_read and cache_creation on every request and split the hit-rate gap into three causes: non-deterministic prefixes, TTL expiry, and sub-threshold blocks.
When Extended Thinking Flattened My Accuracy but Doubled the Bill — Field Notes on Measuring the Marginal Utility of Thinking Tokens
You pinned budget_tokens high 'to be safe,' accuracy barely moved, and the bill kept climbing. These field notes show how to ledger real thinking-token usage by p50/p95/hit-rate and measure how much accuracy each extra 1,000 thinking tokens actually buys.
Tightening Tool Schemas From the Arguments You See in Production
Record the arguments Claude actually passes to your tools in production, then use that distribution to add enums and patterns back into your JSON Schema. With logging code and before/after numbers.
Make Your Nightly MCP Connectors' Health Visible — A Lightweight Ledger for Solo Operators
You don't need Enterprise connector observability to see your MCP connectors' error rate and latency. Append one line per tool call, roll it up weekly, and let regressions ring a bell. A working health ledger for anyone running scheduled tasks solo.
Same Output, Different Path — Guarding Agent Trajectories with Invariants
When the default model changes, your final output can stay correct while the path your agent takes quietly shifts. Here is a trajectory regression harness built on recorded tool traces and deterministic invariants, with working code and measured numbers.
A connector failed for two nights and I never noticed — instrumenting my solo setup after observability went to public beta
The week connector observability hit public beta, I realized my one-person operation had no view into errors or latency. Here is how I wrapped my MCP connector calls in a thin meter and started reviewing it weekly.