Guarding Gemini API Responses in CI: Snapshot and Semantic Regression Testing
Defend non-deterministic Gemini API responses with pytest snapshots plus embedding-based semantic regression detection: record/replay CI, a weekly live smoke that keeps cassettes honest, per-field drift reporting, threshold calibration, and update governance — all in working code.
Keeping Apps Script + Gemini Automations on Least Privilege: Explicit Scopes and Catching Scope Creep
Apps Script automations that call Gemini quietly accumulate OAuth scopes. Here is how to declare explicit scopes in appsscript.json, catch scope creep in CI, and avoid forcing every user to re-consent.
Read Video with Timestamps in the Gemini API: Pull Just the Scene You Need
Hunting for 'where was that step?' in a screen recording or app demo is a chore. Here is how to use Gemini API video understanding to pull just the right scene with timestamps, plus a design that keeps tokens down with FPS and resolution.
The Morning a Managed Agent Stalled and Left No Trace — Building a Run-Observability Layer Outside the Sandbox
With Gemini Managed Agents, the sandbox lives on Google's side, so when a run stalls there is nothing left in your own logging stack. This is a working TypeScript design for an outside observability layer that taps stream events into a ledger, detects silent stalls, and folds runs into readable postmortems.
When Gemini × Qdrant Hybrid Search Was Quietly Losing Recall — Field Notes on Instrumenting RRF Weights and Sparse-Vector Drift
Run Gemini embeddings with Qdrant hybrid search in production and your dashboards stay green while recall quietly slips. These field notes show how to catch it with measurement — RRF weights, sparse-vector drift, missing payload indexes — and protect it with a quality budget.
Audit Your Gemini API Key Restrictions Before the Unrestricted-Key Block Stops Your Automation
Since June 19, 2026, requests from unrestricted Gemini API keys are blocked. Here is how to check whether your key is affected and add restrictions without taking your production automation down.
Stopping Runaway Costs Twice: Project Spend Caps Plus an App-Side Soft Limit
Pairing Gemini API Project Spend Caps (a monthly USD ceiling) with an app-side soft circuit breaker that trips before the hard cap. Includes a working Python and sqlite daily cost ledger.
Your Gemini Structured Output Keys Keep Reordering — Pin Them With propertyOrdering
You constrained the shape with responseSchema, yet the JSON key order shifts between calls and your snapshot tests go red for no reason. Here is why field order is not guaranteed by default, how propertyOrdering fixes it, how Pydantic sets it for you, and how to align few-shot examples — all with working code.
When your Gemini API spend cap trips, paying users go down too — isolating the blast radius with per-tier projects
A Project Spend Cap stops the entire project at once. To keep a runaway free tier from taking paying users down with it, this is a design note on isolating the cap's blast radius across per-tier projects and closing the ~10-minute delay with an application-side soft budget gate.
Building an Intelligent Data Layer: Gemini API × GraphQL Architecture Guide
A comprehensive guide to integrating Gemini API into GraphQL for AI-powered resolvers, semantic caching, and type-safe schema design. Build production-grade intelligent data layers.
When a Deploy Drops the Webhook: Reconciling Gemini Long-Running Operations with a Belt-and-Suspenders Design
Even after you move from polling to Webhooks, events still get dropped during deploys and transient 5xx windows. Here is how I double up Gemini long-running operations with an operation ledger and a low-frequency reconciliation poller so a missing terminal event never goes unnoticed.
Your Gemini API Average Latency Looks Great — But Some Users Still Get Stuck. Defending p95/p99
Your average TTFT is fast, yet a fraction of users keep hitting frozen responses. That is a tail-latency problem (p95/p99). From measurement to model routing, streaming budgets, cache accounting, and retry design — here are the defenses that actually held up in production, with code.