AI Agent Workflow Collection

Calculate the Software Carbon Intensity (SCI, ISO/IEC 21031:2024) of the current project — the CO2 emitted per unit of useful work. Use when the user asks to measure, estimate, compute, or track the carbon footprint / SCI / carbon intensity of this codebase, a service, an endpoint, or a feature, or to compare emissions before and after an optimization.

Apply the Green IT "Carbon Awareness" principle — the same kWh emits very different CO2 depending on when and where it runs (grid mix varies by factor ~15–30). Use when choosing a cloud region, writing a cron/batch job, backup, index rebuild, ML training run, or any time-flexible workload; when deferring non-urgent network sync on a device; or when adapting behavior to connection/data-saver state.

Apply the Green IT "Algorithmic Efficiency" principle — every CPU cycle costs energy, so minimize and avoid computation, especially on high-frequency paths. Use when choosing an algorithm or data structure, profiling a hot path, rendering long lists, deciding SSR/SSG vs client rendering, adding caching, picking a runtime for CPU-bound work, or optimizing a query or render loop.

Apply the Green IT "Longevity" principle — embodied (manufacturing) carbon dominates a device's lifetime footprint, so extending hardware and software lifetime is one of the strongest levers. Use when setting performance budgets, targeting older/low-end devices, choosing framework/runtime versions (LTS vs bleeding edge), designing API payloads/versioning, building container images, or planning hardware refresh/procurement.

Find the highest-impact Green IT levers in a project before optimizing (Pareto / focus-on-impact). Use when the user asks where to start on energy/carbon, what the biggest levers are, how to reduce a project's footprint, or to prioritize a green optimization effort. Counters the bias of tuning interesting-but-marginal code while ignoring bigger structural waste.

Apply the Green IT "technological understanding" principle — software runs on physical hardware with properties you can exploit for free efficiency gains. Use when choosing UI colors / dark mode or CSS animations, selecting a CPU architecture (ARM vs x86) or instance type, configuring TLS/crypto libraries, deciding hot/warm/cold data placement (RAM vs disk), or tuning BIOS/cooling on self-managed hardware.

Apply the Green IT "Less is more" principle — the most effective energy measure is not building, storing, or running something in the first place. Use when adding a feature, deciding what data to persist, reviewing scope, wiring up a web font or third-party widget, cutting an interaction flow, setting log levels, choosing service granularity, or deciding whether an environment/service needs to run 24/7.

Apply the Green IT "Hardware Efficiency" principle — idle resources waste 30–50% of peak power for no work, so maximize utilization and avoid idle draw. Use when sizing cloud instances, configuring autoscaling or serverless/scale-to-zero, designing for load spikes (queues), reviewing memory leaks or event listeners, replacing polling with observers/push, or setting Kubernetes resource requests / bin-packing.

Apply the Green IT "Network Efficiency" principle — every transferred and every stored byte costs energy across server, network, and device. Use when handling images/media, JS bundles or imports, writing database queries or ORM code, designing APIs/response payloads, choosing a protocol (HTTP/1.1 vs HTTP/2 vs gRPC), setting up polling, or configuring compression, CDNs, caching, storage tiers, retention, or backups.