What problem does it solve? Engineering decisions often rely on guesswork, popularity signals, or surface-level documentation. This Skill enforces an evidence standard for technical research: it traces how systems actually work in source code and tests, compares alternatives against local constraints, and produces a recommendation with labeled confidence levels. ## Core Features & Use Cases - Scoped research depths: Choose Quick, Standard, or Deep investigation based on risk, avoiding unnecessary landscape surveys. - Source deep dives: Trace entry points, state flow, failure paths, and tests in local or external repositories using the deep-dive rubric in references/deep-dive-guide.md. - Evidence discipline: Label claims as direct evidence, inference, or unknown, with file and line citations pinned to commits or tags. - Use Case: When choosing between Temporal, Inngest, and Trigger.dev for durable workflow execution, the Skill first inspects your existing job state and tests, then drills into each candidate's persistence and retry mechanisms, and returns an adopt/adapt/experiment/reject recommendation with remaining validation steps. ## Quick Start Ask the agent to compare how two open-source libraries implement a mechanism you need and recommend which approach to adopt for your project.
