Skip to main content

Source: docs/adr/0023-rename-framework-from-rundown-to-lernkit.md

0023 — Rename the framework from Rundown to Lernkit

Context and Problem Statement​

The project's working name was Rundown. Before committing code publicly we ran the trademark / common-law / domain-availability check originally scheduled for P0 exit (OQ-P1-2). The findings — summarized below — made Rundown a bad-value name to ship under. This ADR records the rebrand to Lernkit, the evidence that supports it, and the scope of downstream changes that were made in-repo on the same day.

Decision Drivers​

Findings that forced the rebrand​

  • DPMA (Germany) mark 306647095 for word mark "Rundown" (classes 9 / 38 / 42) — dead since 2016-10-31 by owner-requested deletion. Not itself a block, but the starting point for the investigation.
  • EUIPO — no indexable bare-RUNDOWN EUTM in classes 9 / 41 / 42. Authoritative eSearch and TMview are JS-gated; confidence medium.
  • USPTO — no live RUNDOWN registration in classes 9, 41, or 42; one live class-25 (apparel) registration of bare RUNDOWN held by Christian Scott; several RIG RUNDOWN (class 9 / 41 — narrow, string-instrument-specific).
  • UKIPO + WIPO Madrid — no indexable bare-RUNDOWN hits in our classes. Confidence low-medium; CAPTCHA-gated.
  • Common-law (the decisive evidence).
    • The Rundown AI / Rundown Media Inc. (rundown.ai, ≈ 2M newsletter subscribers, paid "AI University" with an "AI for Coding" course). Senior US common-law user directly in our target class 41 (education services) and class 9 (downloadable educational software). A class-41 registration by a new "Rundown" product would likely draw an opposition; even without an opposition, SEO and brand-confusion cost is material.
    • elseano/rundown — a live GitHub OSS developer CLI that runs executable Markdown. Same category as Lernkit, same bare name, not a legal block but an immediate naming and discovery collision on day one.
    • Rundown Studio, Rundown Creator, Rundown Assistant — active class-9 broadcast / production SaaS products named "Rundown".
    • TheRundown.io — sports-odds API (class 42 overlap).
    • Multiple "The Rundown" media brands (NBC, NEPM, Public.com) in class 38 / 41.
  • Domains — rundown.ai, rundown.io, rundown.studio, rundownstudio.app, rundown.com, rundown.dev all held. Only long-tail or hyphenated variants available.

The decision criterion​

Even though the registered-mark picture in classes 9 / 41 / 42 was actually clear of direct federal blockers, three practical factors made "Rundown" a bad-value fight:

  1. A strong US common-law senior user (The Rundown AI) in the most relevant class (41 — education services).
  2. A same-category OSS project with the identical name on GitHub (elseano/rundown).
  3. Every usable .dev / .io / .com / .ai / .studio domain already taken.

The cost of the rebrand at Phase 0 (no public code, no npm publish, no GitHub org, zero external references) is near-zero; the cost of the rebrand at any later phase compounds sharply.

Considered Options​

  • A: Rebrand to Lernkit (shortlisted + cleared against a pre-check of TLDs, npm, GitHub; full trademark search still needed at P0 exit but with Lernkit as the target).
  • B: Keep "Rundown" but register only in classes 9 + 42 (skip 41 to avoid The Rundown AI opposition); use a hyphenated domain; accept the SEO headwind.
  • C: Keep "Rundown" and file in all three classes anyway; respond to any opposition if and when it comes.

Decision Outcome​

Chosen option: A — rebrand to Lernkit.

Why Lernkit passed the pre-check​

Pre-check across eight candidates (Didax, Kadex, Kelm, Codrum, Scriptum, Forma, Vade, Lernkit) on 2026-04-22:

  • All six relevant TLDs unregistered — lernkit.dev, lernkit.io, lernkit.com, lernkit.app, lernkit.sh, lernkit.ai returned no A record and no SOA. The only candidate that cleared every TLD.
  • Unscoped npm package lernkit available on the npm registry.
  • GitHub org / user lernkit available.
  • Semantics — German compound "lern" (learn) + "kit" reads cleanly as "learning kit", which matches the product (OSS single-tenant framework for producing training courses) and stays neutral about the runnable-code angle so it does not over-commit to a feature.
  • Didax was eliminated despite a clean TLD sweep: didax.com resolves to Didax, Inc., a live US K-12 educational-products publisher — a direct class-41 adjacency.

Scope of the rebrand in this repo​

Performed on 2026-04-22 in the same commit:

  • File renames: rundown.config.ts → lernkit.config.ts; apps/api/src/rundown_api/ → apps/api/src/lernkit_api/; apps/docs/src/styles/rundown.css → apps/docs/src/styles/lernkit.css; infra/coolify/rundown.yml → infra/coolify/lernkit.yml.
  • Content substitutions across every source file: RUNDOWN → LERNKIT, Rundown → Lernkit, rundown → lernkit, rundown-dev/rundown → manykarim/lernkit. Lockfiles regenerated (pnpm-lock.yaml, apps/api/uv.lock).
  • npm scope: @rundown/* → @lernkit/*.
  • Docker Compose project name: rundown-dev → lernkit-dev.
  • Python package name: rundown-api → lernkit-api; Python module directory: rundown_api → lernkit_api.
  • Env var prefix: RUNDOWN_* → LERNKIT_*.
  • Documentation references to the product name across all ADRs, DDD docs, and plan docs.

The one file deliberately left alone is the research document at docs/research/compass_artifact_*.md — it predates naming and contains zero occurrences of either name.

Consequences​

  • Clarity, good: the product name matches the on-disk package names, the on-GitHub repo name, and the as-published npm scope. Single source of truth.
  • Portability, good: a clean first commit under the final name avoids a painful rename across downstream references (mentions in third-party blogs, bookmarked commits, etc.) later.
  • Adoption, mixed: "Lernkit" is a German compound — immediately readable to German speakers, recognizable as "learn kit" to English speakers, but less phonetically memorable than a single-word name. Net-positive for the target audience (technical trainers, including a meaningful DACH-region slice via the Robot Framework and testing-automation communities).
  • Trademark, improved: Lernkit is still subject to the full paid legal search at P0 exit (OQ-P1-2 remains a valid open question, retargeted at Lernkit). The pre-check only confirmed no indexed collisions on domains / npm / GitHub and no common-law uses reachable via web search.
  • Cost, near-zero: the rebrand happens before any public code, npm publish, or GitHub org reservation. No downstream breakage.

Pros and Cons of the Options​

A — rebrand to Lernkit — chosen​

  • Good: clean domains, clean npm, clean GitHub; no common-law collision surfaced in web search.
  • Good: reads as a compound with clear meaning (learn + kit) in both English and German.
  • Good: the Phase 0 cost of rebranding is the smallest it will ever be.
  • Bad: loses the "runnable code" hint that "Rundown" carried phonetically. Accepted; the three-letter "run" is not a strong enough brand hook to keep a landmined name for.

B — keep "Rundown" in classes 9 + 42 only; use a hyphenated domain​

  • Good: no rebrand cost today.
  • Bad: compromised domains and SEO indefinitely; The Rundown AI still wins "rundown" brand searches; elseano/rundown still creates package / repo naming confusion on GitHub.

C — keep "Rundown" and file in class 41 anyway​

  • Good: simplest today.
  • Bad: materially raises opposition risk against The Rundown AI's common-law senior use in exactly that class.

Validation​

  • Pre-check (2026-04-22) — eight candidates batch-checked against DNS SOA records for six TLDs, the npm registry, and the GitHub users API. Only Lernkit swept all three.
  • Rebrand completeness — grep -rI "[Rr]undown\|RUNDOWN" across the repo (excluding generated outputs and lockfiles) returns zero matches.
  • Post-rebrand validation pipeline, 2026-04-22:
    • pnpm lint — 14 files, 0 errors.
    • pnpm test — 9 / 9 vitest tests pass.
    • pnpm build — 3 / 3 packages build; 6 HTML pages produced; Pagefind index intact.
    • uv run ruff check + uv run ruff format --check — all pass.
    • uv run mypy src — 0 issues in 5 source files.
    • uv run pytest — 3 / 3 pass.
  • Full paid trademark search — still required at P0 exit (2026-05-11), now retargeted at Lernkit (OQ-P1-2).

Open questions this ADR updates​

  • OQ-P1-1 (npm scope): decided value changes from @rundown/* to @lernkit/*. Reservation at P0 exit still in force.
  • OQ-P1-2 (trademark search): the provisional clearance of Rundown via the no-paid-search route is withdrawn; the commissioned search at P0 exit targets Lernkit instead. Pre-check evidence noted above suggests Lernkit will clear the paid search, but the paid search is still required.

More Information​

  • Research §9 "Open questions worth resolving before phase 1".
  • Trademark search agents' findings, 2026-04-22 (EUIPO, USPTO, UKIPO + WIPO Madrid, common-law).
  • Pre-check results for the candidate shortlist, 2026-04-22.
  • ADR 0014 — MIT license (unchanged).
  • ADR 0022 — OSS single-tenant scope (unchanged).
  • docs/plan/10-open-questions.md — decision log updated with OQ-P1-1 and OQ-P1-2 resolutions dated 2026-04-22.