feat: add React Router adapter

[farizanjum]

Closes #6

Summary

• add @dualmark/remix for React Router v7 Framework Mode
• generate markdown resource routes and /llms.txt from Dualmark config
• add entry-server negotiation for bot user agents and Accept: text/markdown
• add examples/remix-blog plus docs and release metadata

Testing

• bun run --filter @dualmark/remix test
• bun run --filter @dualmark/remix typecheck
• bun run --filter @dualmark/remix build
• bun run --filter dualmark-example-remix-blog build
• bun run --filter dualmark-docs build
• node packages/cli/dist/cli.js verify http://localhost:5174/posts/hello returned 125/125 under react-router dev
• node packages/cli/dist/cli.js verify http://localhost:3000/posts/hello returned 125/125 under react-router-serve

Notes

• slugStrategy: "catch-all" is intentionally rejected for now because React Router splats cannot safely express nested *.md routes without catching human HTML paths.
• The current React Router docs entry point is https://reactrouter.com/start/framework/installation.

[vercel]

@farizanjum is attempting to deploy a commit to the Dodo Payments Team on Vercel.

A member of the Team first needs to authorize it.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[dodo-squirrels]

Review: feat: add React Router adapter (@dualmark/remix)

Verdict: Approve with follow-ups. Strong, well-structured feature. Two spec-consistency nits below that I'd like to see addressed (ideally before merge, since they re-introduce bugs that sibling PRs #70/#73 are fixing) — neither blocks the core functionality.

What I exercised

Checked out the branch, ran a clean bun install, then:

• @dualmark/remix: build ✓, typecheck ✓, 24/24 tests ✓
• examples/remix-blog: builds, generates the 4 markdown resource routes + llms.txt
• Ran the production server (react-router-serve) and drove it end-to-end with curl
• Ran the CLI conformance verifier against the live server

Note: on first run test/typecheck failed with Cannot find module @dualmark/core — that was a stale-lockfile linking artifact in my sandbox; a fresh bun install fixed it. Not a PR defect.

Independent conformance result

node packages/cli/dist/cli.js verify http://127.0.0.1:3000/posts/hello
Score: 125/125   ← all 14 checks pass (incl. 406 negotiation, Link alternate, tokens, noindex, nosniff, aeoVersion)

End-to-end header checks all correct: browser → text/html + Link rel=alternate + Vary; GPTBot → text/markdown + X-Robots-Tag: noindex + X-Markdown-Tokens; Accept: text/markdown → markdown; direct .md → markdown; /llms.txt → text/plain.

Code quality — strong

• config-validation.ts is thorough: rejects trailing-slash siteUrl, non-relative/.. paths, and correctly rejects Windows absolute paths cross-platform via win32.isAbsolute (the follow-up commit). Route/pattern sanitization is solid.
• routes.ts GENERATED_MARKER guard refusing to overwrite user-authored route files is a nice safety touch.
• handlers.ts targeted dispatch + resource-route factoring is clean and reuses core primitives correctly.
• Intentionally rejecting slugStrategy: "catch-all" with a clear error is the right call for the splat limitation.

Follow-up 1 (spec §5) — bot UA should respect explicit Accept: text/html

entry-server.ts:78 uses the legacy if (bot.isBot || format === "markdown"). I confirmed live: a request with User-Agent: GPTBot/1.0 and Accept: text/html is served text/markdown + noindex on the canonical URL:

curl -H 'User-Agent: GPTBot/1.0' -H 'Accept: text/html' …/posts/hello
→ content-type: text/markdown; x-robots-tag: noindex   ✗ (spec §5: UA MUST NOT override explicit Accept)

PR #70 just introduced shouldServeMarkdown(accept, isBot) in @dualmark/core for exactly this. Please route this adapter through it too.

Follow-up 2 (spec §3) — Vary: Accept gated on injectLinkHeader

entry-server.ts:82-87 only calls appendVaryAccept inside if (injectLinkHeader). With injectLinkHeader: false, negotiable HTML would ship without Vary: Accept — the same cache-poisoning bug PR #73 is fixing across the other adapters. Vary: Accept should be set whenever the page is negotiable, independent of the Link flag.

Minor

• Package/PR is named "React Router" but publishes as @dualmark/remix — intentional given RRv7 = Remix lineage, just calling it out for discoverability (README/docs do explain it).

Overall this is a high-quality adapter that already hits perfect conformance. Approving; please fold in the two @dualmark/core helpers (shouldServeMarkdown, unconditional Vary) so it lands consistent with #70/#73.

[farizanjum]

Thanks for the detailed review. I addressed the Vary follow-up in 3054b11: negotiable HTML now receives Vary: Accept even when injectLinkHeader is false, with a regression test covering Link absent plus Vary present.

For the explicit Accept follow-up, PR #70 is still open and owns the shared shouldServeMarkdown helper. I will rebase onto it once it merges and route @dualmark/remix through that helper rather than duplicate its negotiation logic here.

[farizanjum]

Thanks for the detailed review. Both follow-ups are now addressed.

• Added @dualmark/core shouldServeMarkdown(accept, isBot), consistent with the shared negotiation approach in fix: respect explicit Accept: text/html from AI bot user agents #70.
• Updated @dualmark/remix to use it, so GPTBot plus Accept: text/html now receives HTML with no noindex header.
• Kept Vary: Accept independent of injectLinkHeader, with a regression test for Link disabled plus Vary present.
• Added core and Remix regression coverage.

Verification:

• bun run test: 26 Turbo tasks successful
• bun run typecheck: 27 Turbo tasks successful
• Remix dev and production conformance: 125/125
• GPTBot plus Accept: text/html: text/html with no X-Robots-Tag