Run one-on-ones the report owns, splitting career growth from status, with notes that compound. Use when establishing 1:1 practice as a manager or getting more from 1:1s as a…
Model UI as explicit finite states so impossible combinations cannot render. Use when a component juggles interacting booleans, shows contradictory states, or a flow has grown too…
Reason about the JavaScript event loop, microtasks vs macrotasks, and why blocking it freezes everything.
Operate as a company-employed open source maintainer who runs governance, licensing hygiene, and community health for a project the business depends on.
Meter consumption accurately enough to bill on it, with reconciliation, limits, and transparency the customer can verify.
Operate as a cloud architect who lays the landing zone, governs cost, and sequences migration for an organization.
Move between API versions or providers without downtime, using parallel running and staged cutover. Use when a provider deprecates a version or you are replacing a service.
Tune gradient-boosted trees (XGBoost, LightGBM, CatBoost) effectively: the parameters that matter and the order to tune them.
Run continuous compliance with agents that map controls to evidence, collect it on a schedule, and flag drift before an audit finds it.
Decide between patching upstream, forking, and vendoring a dependency, and carry the ongoing cost knowingly.
Design each handoff between two agents as an explicit artifact contract with a context budget and a rule for what must survive the crossing.
Place data across the GPU memory tiers so a kernel is limited by math rather than by repeated trips to HBM.
Order material so each piece is learnable when it arrives, with prerequisites satisfied and difficulty increasing gradually.
Operate as a UX researcher who ties every study to a pending decision, synthesizes evidence into ranked insights, and delivers findings that move the roadmap.
Record an execution once and replay it deterministically so you can step backward to the moment a value went wrong.
Write React hooks with correct dependencies and minimal effects so renders stay predictable. Use when adding useEffect, debugging stale closures or render loops, or extracting a…
Get teams onto the supported path through incentives and migration support rather than mandates. Use when a better platform exists and teams are not moving to it.
Run an author agent and a reviewer agent in a loop with a strict finding format, a mandatory verification pass, and a convergence rule that ends it.
Build disaster recovery around RTO/RPO tiers, verified backups, and drills that prove the numbers. Use when writing a DR plan or testing whether the existing one actually works.
Write benchmarks that do not lie, controlling warmup, isolation, and statistics so the number reflects the code and not the environment.
Quantize a trained model to INT8 or INT4 for inference, calibrate the ranges, and gate the release on a measured quality regression.
Decide where a web app keeps tokens and state by reasoning about the XSS blast radius of each store, not by convenience.
Build a modular monolith with clean internal seams and split to services only on proven pressure. Use when starting a system or resisting premature microservice decomposition.
Process only new and changed data with watermarks, merge strategies, and a full-refresh escape hatch.
Understand address translation, paging, and the memory hierarchy well enough to explain swap thrashing, page faults, and why memory usage numbers disagree.
Let clients work disconnected and reconcile changes on reconnect without losing edits or resurrecting deletions.
Write SQL other people can review, modify, and trust, with clear structure, meaningful aliases, and stated assumptions.
Trust your cross-validation over the public leaderboard, avoid overfitting it, and select final submissions wisely.
Cut writing to its essential words so every one earns its place, without losing meaning or voice. Use when a piece is too long, dense, or padded, or whenever brevity respects the…
Place human approval at the points where being wrong is expensive or irreversible, with enough context to decide quickly. Use when agents perform work that has real consequences.
Format code on the current branch using Biome. Use when asked to format, lint, or clean up code before committing or creating a PR.
Prepare documents for physical printing with correct colour, bleed, resolution, and fonts, so what prints matches what you designed.
Add type hints where they catch real errors and document contracts, choosing Protocol, generics, or TypedDict deliberately and setting a checker strictness you can actually hold.
Know what the company can actually take on this cycle, in agent budget and human hours, and refuse work beyond it. Use when everything is committed and nothing finishes.
Decide between a provider's SDK and direct HTTP calls, weighing convenience against dependency weight and control. Use when starting an integration.
Define who decides what in a project, how that changes, and how disputes end, before a disagreement forces the question.
Classify incidents consistently so the response, communication, and escalation match the impact. Use when incident response is either overwhelming or too slow.
Keep running discovery rounds until several consecutive rounds find nothing new, rather than stopping at a fixed count.
Use ARIA correctly by preferring native HTML, applying roles/states/properties only where needed, and following known widget patterns.
Stop trusting wall clocks for ordering and expiry across machines; use logical clocks and single-writer timestamps.
Group and aggregate correctly, understanding what GROUP BY collapses, how HAVING differs from WHERE, and how nulls and empty groups behave.
Choose which browsers, versions, and devices to test from real usage data instead of habit or a vendor's full grid.
Choose between operational transformation, CRDTs, and locking for multi-user editing, and understand what each costs.
Assert hard-to-specify output by reviewing a human-readable snapshot once, approving it, then failing on any later diff.
Turn on and migrate to TypeScript's strict flags incrementally so the compiler catches real bugs. Use when configuring TypeScript strictness or tightening a loose codebase without…
Write Kotlin that uses null safety, data classes, and coroutines as designed, with clean Java interop. Use when writing Kotlin or migrating Java code and habits to it.
Operate as a platform engineer who builds golden paths for internal developers and earns adoption on merit.
Design MCP tools an agent can select correctly, with names, descriptions, and schemas that make the right call obvious and the wrong one impossible.
Model warehouse tables with explicit grain, conformed dimensions, and deliberate SCD handling. Use when designing analytics schemas or fixing double-counted metrics and unjoinable…
Design React component props and boundaries so components compose cleanly and stay reusable. Use when creating a component, reviewing a bloated props list, or deciding controlled…
Staff a pipeline with agents that research accounts, qualify leads against explicit criteria, draft outreach for human sending, and keep the CRM honest.
Write changelogs for humans, grouped by change type, with breaking changes called out and versions dated.
Drive a real debugger with breakpoints, watch expressions, and conditional stops to read live program state at the moment of failure.
Attribute memory growth and allocation churn to the exact call sites that produce it using an allocation profiler.
Judge a canary release against a concurrent baseline using pre-set metric gates, automatic rollback, and statistically honest windows.
Stop UI redress attacks by declaring who may frame your pages, using frame-ancestors as the primary control and X-Frame-Options as the fallback.
Structure game loops with fixed-timestep simulation, interpolated rendering, and spiral-of-death protection.
Read code metrics as smoke that points at where to look, not as verdicts that rank code good or bad. Use when triaging a large codebase for refactoring targets or reviewing…
Draw honest conclusions from samples using hypothesis tests, confidence intervals, and significance read correctly.
Implement OAuth 2.0 and OpenID Connect flows that bind the request end to end with PKCE, state, and exact redirect allowlists.