Design · Checked 12 August 2026

Flashback: Give an AI design task a specific year

Turn design history into recipes, constraints, and anti-clichés.

Flashback is an agent skill and design-history corpus that turns a requested year into concrete visual directions and guardrails.

Type
skill
Works on
Agent skills, GitHub Pages

Open Flashback ↗︎

Design history as working material

Flashback is an agent skill backed by a year-by-year design corpus. For a chosen year, it gives the agent a thesis, historical signals, typography and graphic language, named recipes, reference artefacts, and clichés to avoid. The skill then turns that research into a brief, system, critique, handoff, or set of competing visual directions.

The useful distinction is specificity. “Make it look eighties” tends to produce the same neon costume. Flashback asks the agent to fetch the actual year, choose a recipe that fits the job, explain which historical traits it is carrying forward, and respect the corpus’s anti-clichés.

We applied its 1981 “mutable broadcast identity” recipe to a real Swipe social card. The result kept the black brand panel stable while the issue graphics used the recipe’s electric blue, hot pink, acid green, block type, and changing broadcast-frame structure. The anti-clichés kept it away from generic vaporwave, cyberpunk, and late-eighties nostalgia.

Use it on the final format

Pick a year that matters to the product or audience, then ask the agent for two different territories. Require the historical references, chosen recipe, and anti-clichés for each direction before generating a design. Apply the selected direction to a real page, ad, or social card at its final size; a large moodboard can hide weak typography and contrast.

The skill fetches its manifest and corpus from the public Flashback site. If a year is missing, it uses the nearest available year and should say so explicitly. The repository is MIT licensed and does not require a hosted account.

A direction, not an authority

Flashback supplies creative constraints rather than verified historical truth. The corpus was assembled with AI assistance, source depth varies by year, and a plausible recipe can still flatten a complicated period into one visual story. Check important historical claims against primary sources when accuracy is part of the deliverable.

The output also depends on the host agent’s design ability. A good recipe does not guarantee legible type, accessible contrast, or a coherent final layout. Review the result at its intended size and treat the historical direction as a starting point for critique.

Featured in Swipe

Sources

Swipe the best AI skills, prompts, and workflows.

The newsletter to learn AI by stealing the cool ideas, skills, and tools you didn't know you needed.

One more step.

Check your inbox to confirm your subscription. Can't find it? Check your spam or junk folder.