Tech Research Workflow

Evidence-driven technology and competitor research that ends in a decision-ready report family, not a pile of links. Use when the question is design-shaping — "should we support X, and how" — and the answer must survive a technical review.

Prerequisite check: this skill declares effective-web-research in frontmatter dependencies. On load, verify it is available; if missing, abort and print the install command (npx skills add FuDesign2008/open-skills -g). No silent fallback — every external search in this workflow runs under that skill's discipline.

Core principle

A design is a function of two inputs: your own business × competitor first-hand evidence.

• Trends and articles answer "should we care about this" — they are the backdrop, never the design input.
• Competitors who shipped the feature already paid the tuition in production: their pits, costs, and incident reports are free intelligence. Only they answer "how, and what does it cost".
• First-hand evidence beats second-hand narrative at every layer: run the shipped feature yourself, read your own code with file:line in hand, and treat every article as a lead to verify.

The order is load-bearing: audit yourself first, research competitors second, design third. Skipping ahead to design is the classic failure mode of this genre — a trend piece arrives, the agent slides straight into solutioning, and the review surfaces wrong assumptions about scope, stance, and constraints that a one-day audit would have caught.

Scope and path selection