Skip to content

awesome-use-case-model.md

Latest commit

 

History

History
86 lines (66 loc) · 4.83 KB

awesome-use-case-model.md

File metadata and controls

86 lines (66 loc) · 4.83 KB
Raw

Awesome Use Case — FINAL model & blueprint

Status: FINAL / approved (2026-08-21). This is the record of truth for how every Awesome Use Case listing is created — by us now, and by autonomous sourcing agents later. Supersedes the old "Bot · Role name + one-line tagline + single category" model.

The card (what a listing shows)

Five parts, each justifiable:

  1. Score eyebrow — the Awesome Score (0–100) + tier word. Top of the card.
  2. Hook (headline) — the exciting outcome/claim. Not a role label.
  3. Summary (summary) — 2–3 lines, what it does + the twist.
  4. Categories — 1–3 neutral chips (multi). The only taxonomy.
  5. Source — via @handle · 𝕏 or via <Channel> · YouTube. No bot name.

Removed vs. old: bot name / "Bot · Role", tags, the metrics strip (setup/difficulty/etc.).

Fields

Required: headline, summary, categories[], awesome_score + score_breakdown, format, source (X post / YouTube video + author handle or channel), slug, added_at, updated_at, status.

Optional: prompt_provenance (author|curator), replicability, author credit, featured.

Kept in data, off the card for now (may return if defined crisply): setup_minutes, difficulty, autonomy, schedule.

format: use-case (scored, actionable) or guide (a reference/explainer — badged "reference", not scored).

Field rules (agent-executable)

  • headline — ≤60 chars, sentence case, no trailing period. Lead with the outcome or the surprising claim, concrete. Use the source's own vivid framing when strong; never add a claim the source doesn't make. No generic role labels. Unique across the directory. Banned slop: supercharge, unleash, seamless, effortless, game-changer, revolutionary, 10x (unless literally in the source).
  • summary — 140–240 chars, 1–2 sentences. What the agent does + the specific twist. Concrete nouns (tools, roles, cadence, artifact). Any number must come from the source/artifact.
  • categories — 1–3 from the taxonomy; every domain it genuinely spans, no padding; most central first. Taxonomy only; new categories are human-approved additions made only when a real cluster needs them. Current 10: work, sales, marketing, engineering, support, finance-ops, data, personal, trading-crypto, fun. Watch-list for additions: operations, research, creative.
  • source + credit — X → via @handle · 𝕏; YouTube → via <Channel> · YouTube. Store exact URL + timestamp; the live embed renders on the detail page. Credit is dofollow once verified.
  • awesome_score — the six-dimension rubric below; store the breakdown; recompute on change; zero input from engagement or author size.
  • integrity (non-negotiable) — every entry traces to a real source; no fabricated tools, quotes, outcomes, or numbers. Author's prompt shown → captured verbatim (prompt_provenance: author, byte-identical). Reconstructed → curator, and the page says so. verified_at only after source + claims are checked.

The Awesome Score (quality, never popularity)

Dimension Max Top of range
Reproducibility 25 verbatim author prompt OR complete copy-paste setup + steps (25) · solid curator prompt + clear steps (17) · conceptual only (8)
Ambition 20 whole business/department/team (20) · real multi-step workflow (13) · single task (7)
Concreteness 20 names specific tools/roles/cadence/artifact (20) · specific but light (13) · generic (6)
Novelty 15 genuinely new pattern (15) · fresh take on a known one (9) · near-duplicate (4)
Evidence 10 inspectable artifact — repo/skill/video demo (10) · screenshot of the real setup (6) · claim only (3)
Craft 10 setup + prompt + why all strong & clear (10) · mostly (6) · thin (3)

Bands: 90–100 must-try · 78–89 awesome · 65–77 solid · 50–64 review · <50 skip.

Uses: (1) the eyebrow badge + default sort; (2) featuring — ≥90 are home candidates; (3) autonomous publish gate — agents auto-publish ≥65, queue 50–64 for human review, discard <50.

Detail page order

Score eyebrow + hook + summary → categories + verified credit → the live source embed → How it's set up → The Prompt (verbatim or curator-labelled) → Why it's cool → Replicability → related (via shared categories).

Pipeline (for autonomous sourcing agents)

  1. Source — real X post / YouTube video / public repo. Capture URL, handle/channel, platform, timestamp, any prompt shown (verbatim).
  2. Extract — core claim, concrete setup, prompt, artifact link, domains touched.
  3. Compose — write fields to the rules above.
  4. Score — grade the six dimensions → awesome_score + breakdown.
  5. Gate — ≥65 publish · 50–64 human review · <50 discard. Never publish without a real source and a passed integrity check.