Repository navigation
docs(roadmap): refresh the competitive benchmark and correct the AI-gap premise - #1284
Merged
Merged
Conversation
…ap premise The benchmark was dated 2026-09-19 and covered three competitors, while competitive-landscape.md still carried its March 2026 premise that no platform had shipped LLM-powered Commander deckbuilding. Several have: ManaTap, Farseek, MTG Master, ManaForge and a long tail of smaller sites, most with a free entry tier. The roadmap was being read against a claim that no longer holds. Record what is still defensible instead: the LLM never decides legality (I-02), evidence the player produced themselves (I-07, I-09) and printing-level ownership (I-06). Each is already an initiative, so the delivery order does not change. Also record four upstream contract changes that affect shipped integrations: EDHREC replaced Synergy with Lift on 2026-10-01 (a ratio centred on 1, not a percentage), Moxfield now answers programmatic requests behind Cloudflare, Scryfall moved bulk data to gzipped JSONL and tightened per-endpoint limits, and Commander Spellbook v0.84.0 added per-deck and per-card combo endpoints. Search results claimed EDHREC had shipped an AI deck chat. The source shows EDHREChat was an April Fools page from 2023, offline since 2023-04-04; it is recorded as such rather than as a competitor feature. Adds eight open scopes and five non-goals. No roadmap item is marked complete. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
Codex Review Summary: ❌ FailedThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the
 |
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
dev gained its own competitive revision (#1289) while this branch was open, rewriting the same two documents. Rather than interleave two rewrites, the conflicting sections are resolved in favour of dev: its benchmark table maps each competitor to an initiative, it introduces I-14 and I-16 through I-21, and it centralises sources and prices in competitive-landscape.md. All eight scope items from this branch survived outside the conflict zones and are kept. Re-added on top of dev's structure, because dev did not carry them: - ManaTap, MTG Master and ManaForge, the three AI builders dev's section 4 omits, with what each ships and what none of them does. - The EDHREChat correction: it was an April Fools page from 2023, not an EDHREC AI product. - The two upstream contract changes that affect shipped integrations — Scryfall's gzipped JSONL bulk format and tighter per-endpoint limits, and Commander Spellbook v0.84.0's per-deck and per-card combo endpoints. - Two non-goals: do not follow the AI builders into tournament management, custom cards, Arena wildcards or scanning, and do not compete on generation volume or free-tier generosity. Verified on the merged tree: tsc clean, lint clean, 3509 tests passing. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
|
This branch was successfully deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
The competitive benchmark in
docs/product/roadmap.mdwas dated 2026-09-19 and covered three competitors.docs/product/competitive-landscape.mdstill dated from March 2026 and carried the strategic premise that no platform had shipped LLM-powered Commander deckbuilding. That premise is no longer true, and the roadmap was being read against it.Several upstream contracts we already depend on also changed during 2026 without the docs recording it.
What changed
The AI gap closed — premise corrected
Verified against the products' own pages on 2026-11-07:
"AI-powered" is now table stakes with a free tier. The three claims that remain ours are already initiatives, not new bets: the LLM never decides legality (I-02), evidence the player produced themselves (I-07, I-09), and printing-level ownership and acquisition (I-06). Positioning emphasis moves from "AI builds your deck" to "AI you can check".
Checked and rejected: search results suggested EDHREC had shipped an AI deck chat. Fetching the source shows
EDHREChatwas an April Fools page from 2023-04-01, taken offline on 2023-04-04. It is recorded as such rather than as a competitor feature.Upstream contract changes now recorded
/commanders,/combosand/tags(2026-10-01). Lift is a ratio centred on 1, not a percentage centred on 0 — any surface treating it as a percentage or ranking by it needs rechecking (I-09).User-Agent/Acceptand serialises behind a 100 ms mutex — fine for a 10 req/s sustained ceiling, not for a stricter per-endpoint search limit. Flagged for human confirmation because scryfall.com blocks automated fetching (I-05).Competitor capabilities absorbed as scope
Eight new open scopes, mapped to existing initiatives: Lift semantics + Spellbook endpoint re-check (I-09), Moxfield import brittleness (I-10), Scryfall policy confirmation (I-05), vendor/category/price-band cost breakdown and explicit proxy state (I-06), visible validator verdict (I-02), dated Game Changers data (I-03).
Five non-goals added, including: do not follow the AI builders into tournament management, custom-card creation, Arena wildcard optimization or camera scanning; and do not compete on generation volume or free-tier generosity.
Scope discipline
Documentation only — two Markdown files, no source changes. Per
CLAUDE.mdthis is a dedicatedchore/docsPR becausedocs/product/roadmap.mdmust not be touched on a feature branch. No roadmap item was marked complete; the revision only adds scope and corrects dated claims.docs/project/changelog.mdanddocs/project/progress.mdare untouched.Verification
npx prettier --checkon both filesnpx tsc --noEmitpnpm lintpnpm testThe three code gates cannot be affected by a Markdown-only diff; they are run to satisfy the mandatory pre-PR gate.
Caveats for review
Competitor rows are built from each product's own public pages. Marketing copy is a claim, not evidence — the benchmark says so explicitly, and each row should be confirmed before it justifies implementation. Sections 2, 6, 7, 8 and 9 of
competitive-landscape.mdstill date from March 2026 and are now marked unverified rather than silently presented as current.E2E gate
SKIP_E2E=1was used for the push: Docker is unavailable in the environment this branch was prepared in, and the diff is Markdown only, so the Playwright suite has nothing to exercise. The three code gates were run in full.Note for a separate PR:
src/components/deck/ExternalRecommendationFreshness.test.tsx:17andsrc/components/deck/AISuggestionsPanel.provenance.test.tsx:57both assert"Observed: Nov 2, 2026, 10:30 AM"for a UTC instant, so they only pass when the machine runs in UTC.TZ=UTC pnpm testgives 3505/3505; without it, those two fail on any developer machine outside UTC. Not fixed here to keep this PR documentation-only.🤖 Generated with Claude Code