diff --git a/.changeset/calm-pages-wait.md b/.changeset/calm-pages-wait.md
new file mode 100644
index 000000000..080366f50
--- /dev/null
+++ b/.changeset/calm-pages-wait.md
@@ -0,0 +1,8 @@
+---
+"kitcn": patch
+---
+
+## Patches
+
+- Fix cRPC infinite queries backed by native Convex pagination loading pages
+  before `fetchNextPage()` is called.
diff --git a/docs/plans/304-native-convex-pagination-in-crpc-infinite-queries.md b/docs/plans/304-native-convex-pagination-in-crpc-infinite-queries.md
new file mode 100644
index 000000000..431e1b1c0
--- /dev/null
+++ b/docs/plans/304-native-convex-pagination-in-crpc-infinite-queries.md
@@ -0,0 +1,462 @@
+# Native Convex pagination in cRPC infinite queries
+
+Objective:
+Resolve discussion #304 native Convex pagination behavior; done when the
+3-of-6 case and next-page state pass focused proof and repo gates; plan
+docs/plans/304-native-convex-pagination-in-crpc-infinite-queries.md.
+
+Flow mode:
+one-shot execution
+
+Goal plan:
+docs/plans/304-native-convex-pagination-in-crpc-infinite-queries.md
+
+Template:
+docs/plans/templates/task.md
+
+Primary template:
+docs/plans/templates/task.md
+
+Applied packs:
+- docs (docs/plans/templates/packs/docs.md)
+- package-api (docs/plans/templates/packs/package-api.md)
+
+Task source:
+- type: public GitHub discussion / behavior report
+- id / link: discussion #304,
+  https://github.com/udecode/kitcn/discussions/304
+- title: Partial migration to Infinite Query breaks pagination
+- acceptance criteria:
+  - A native Convex paginated query with six rows and limit three does not
+    auto-fetch through all rows.
+  - The infinite-query next-page state is true before exhaustion and false
+    after exhaustion.
+  - Partial migration does not require replacing native Convex schema
+    validators or the native paginator with the kitcn ORM.
+
+Timed checkpoint:
+- requested duration: N/A: none requested
+- semantics: N/A: no timed checkpoint
+- initial confidence score: N/A: binary behavior and gate threshold
+- improvement loop: reproduce, fix the owner, rerun every case and repo gates
+- final score / loop closure: N/A: evidence-bounded confidence at handoff
+
+Completion threshold:
+- The native Convex 3-of-6 repro fails before the fix and passes afterward.
+- Exactly two user-visible fetch steps expose 3 then 6 rows, with next-page
+  state true then false.
+- Partial migration remains supported without an ORM/schema rewrite.
+- Focused tests, package build/typecheck/lint, changeset, autoreview, `bun
+  check`, commit, push, task-style PR, discussion sync, and goal-plan check
+  pass.
+- Task closure is legal only when the source-of-truth acceptance criteria are
+  satisfied or explicitly narrowed, required verification evidence is recorded,
+  code-review and release-artifact gates are closed when applicable, verified
+  code changes are committed and PR'd unless explicitly declined or blocked,
+  task-style PR body sync is complete or marked N/A with reason,
+  GitHub issue/PR sync is complete or marked N/A with reason, and
+  `node .agents/skills/autogoal/scripts/check-complete.mjs docs/plans/304-native-convex-pagination-in-crpc-infinite-queries.md` passes.
+
+Verification surface:
+- Focused cRPC React infinite-query regression test at the package owner.
+- `bun --cwd packages/kitcn build`, relevant typecheck, `bun lint:fix`, and
+  final `bun check` in `/Users/zbeyens/git/better-convex`.
+- Source audit of native Convex pagination result handling.
+- Changeset, task-style PR body read-back, and discussion #304 sync.
+- Browser proof is N/A unless investigation discovers a browser-only behavior.
+
+Constraints:
+- Preserve existing user-facing behavior outside the task scope.
+- Prefer the durable ownership boundary over caller-by-caller patches.
+- Verified code changes must be committed and PR'd because the task skill
+  requires that path unless the user explicitly says not to, the work has no
+  local patch, or a real blocker is recorded.
+- The absence of a separate "open a PR" sentence from the user is not a valid
+  N/A reason for verified code-changing task work.
+- A PR created by this task must use the PR #270 emoji task-style PR body
+  contract below, not a generic summary/body from a git helper skill.
+- Do not add broad ceremony when the task is trivial or docs-only.
+- Do not require an ORM or schema-validator migration for cRPC infinite-query
+  pagination.
+- Do not add compatibility parsing at callers when the cRPC adapter owns the
+  native pagination-result translation.
+
+Boundaries:
+- Source of truth: GitHub discussion #304 plus current cRPC infinite-query
+  implementation and migration documentation.
+- Allowed edit scope: `packages/kitcn` cRPC pagination owner and focused
+  React/Solid tests; matching docs only when behavior needs clarification;
+  plan, changeset, generated fixture snapshots, and the stale raw Next fixture
+  boundary required to restore the mandatory runtime gate.
+- Browser surface: N/A: hook/runtime package behavior has an honest automated
+  harness and no visual output.
+- GitHub issue sync: reply to discussion #304 after the PR exists.
+- Non-goals: ORM migration, schema conversion, pagination API redesign,
+  unrelated query behavior, Solid changes unless the shared defect exists
+  there too.
+
+Output budget strategy:
+- Use exact-symbol `rg`, bounded `sed` ranges, and focused test commands.
+  Exclude `node_modules`, cap shell output, and inspect only the owning React,
+  Solid, cRPC, docs, test, and local Convex source files needed to settle the
+  result contract.
+
+Blocked condition:
+- Stop only if the native Convex result contract cannot be obtained from local
+  source/tests after a real harness attempt, required package checks remain
+  broken after the one allowed install-corruption retry, or GitHub access
+  prevents mandatory shipping/sync.
+
+Task state:
+- task_type: bug / runtime compatibility
+- task_complexity: normal non-trivial
+- current_phase: review
+- current_phase_status: complete
+- next_phase: commit / PR / GitHub sync
+- goal_status: active
+
+Current verdict:
+- verdict: valid
+- confidence: 97%
+- next owner: task
+- reason: React and Solid source-level hook repros failed because a plain
+  `splitCursor` created a second subscription; both pass after matching
+  Convex's split predicate.
+
+Implementation readiness:
+- verdict: ready
+- exact owner: shared internal cRPC pagination split policy consumed by React
+  and Solid infinite-query hooks
+- contradiction status: settled by Convex's local source contract: a
+  `splitCursor` is only actionable with split page status or page growth over
+  twice the initial item count
+- source-listed cases complete: yes; three rows below
+
+Pre-solution issue challenge:
+- reporter claim: a cRPC infinite query backed by native Convex pagination
+  auto-fetches a six-row result as 3, 5, then 6 rows and reports
+  `hasNextPage === false` throughout.
+- suggested diagnosis or fix: the reporter's agent attributes the behavior to
+  `splitCursor`; no concrete fix is proposed.
+- repro ladder:
+  - tests / source-level repro: required; smallest honest owner
+  - repo-owned automated browser or integration proof: focused React and Solid
+    hook tests reproduce the package behavior directly
+  - Browser plugin: N/A unless source/test harness is dishonest
+  - screenshot / visual proof: N/A: no visual output
+- reproduction verdict: reproduced in both framework hooks; React query count
+  was 2 instead of 1 and Solid query count was 2 instead of 1
+- validity verdict: valid
+- best long-term fix boundary: one shared internal split predicate consumed by
+  both framework hooks
+- harsh honest feedback: requiring an ORM migration would be the wrong fix;
+  the client ignored Convex's own `pageStatus` contract
+- hard-stop decision: proceed; direct failing repro exists
+
+Completion rule:
+- Do not call `update_goal(status: complete)` while any required checklist item
+  remains unchecked. If an item does not apply, check it and add `N/A: <reason>`.
+- Do not call `update_goal(status: complete)` until every completion threshold
+  above is satisfied, final handoff evidence is recorded, and
+  `node .agents/skills/autogoal/scripts/check-complete.mjs docs/plans/304-native-convex-pagination-in-crpc-infinite-queries.md` passes.
+- Do not create hook state for this goal. This file plus the active goal are the
+  durable state.
+
+Start Gates:
+| Gate | Applies | Evidence |
+|------|---------|----------|
+| Timed checkpoint parsed | no | N/A: none requested |
+| Skill analysis before edits | yes | Requested `task`; loaded `autogoal`; `tdd`, `changeset`, and `autoreview` remain conditional gates |
+| Active goal checked or created | yes | `get_goal` returned none; goal created for this exact plan |
+| Source of truth read before edits | yes | `gh api graphql` read discussion #304 body before repo exploration |
+| GitHub comments and attachments read | yes | GraphQL returned zero comments, no answer, and no attachments |
+| Video transcript evidence required | no | N/A: discussion contains no video or recording |
+| Pre-solution issue challenge required | yes | Public behavior claim; reporter claim and source cases recorded |
+| Reproduction verdict before implementation | yes | React Bun test and Solid Vitest each failed with expected query count 1, received 2 |
+| Repro escalation ladder selected | yes | Focused package test/source harness first; browser N/A unless that layer proves insufficient |
+| Suggested fix reviewed against durable boundary | yes | `splitCursor` is a diagnosis to test, not authority; adapter ownership will be settled first |
+| `docs/solutions` checked for non-trivial existing-code work | yes | Exact pagination/cRPC search found no existing solution for this hook defect |
+| TDD decision before behavior change or bug fix | yes | Loaded `tdd`; React and Solid behavior repros were observed red before each fix |
+| Branch decision for code-changing task | yes | Created `codex/fix-native-convex-pagination` from current `origin/main`; unrelated merged branch not reused |
+| Release artifact decision | yes | Published `kitcn` runtime behavior; patch changeset required |
+| Browser tool decision for browser surface | no | N/A: package hook behavior has no visual surface; reassess if repro needs runtime browser proof |
+| Commit / PR expectation decision | yes | `task` requires commit, push, PR for verified code changes |
+| Task-style PR body decision | yes | PR #270 emoji contract required and read-back with `gh pr view --json body` |
+| GitHub issue sync expectation decision | yes | Reply to discussion #304 after PR creation |
+| Output budget strategy recorded | yes | Exact-symbol bounded reads/tests; no unbounded output |
+| Docs pack selected | yes | Incidental current-state migration/docs guidance may need sync |
+| Docs guidance loaded | no | N/A: no public docs edit; current migration guide already shows native `ctx.db.query(...).paginate(...)` under cRPC |
+| Docs lane selected | yes | Supporting docs lane, not docs-dominant |
+| Target docs and nearest sibling docs read | yes | Read `www/content/docs/migrations/convex.mdx`, React infinite-query docs, and published React skill guidance |
+| Docs style doctrine read | no | N/A: no `www/**` or published skill content change |
+| Documented source owner identified | yes | Runtime owner is package hook/shared predicate; current docs already support partial native pagination |
+| Package/API pack selected | yes | Package runtime compatibility/release behavior is in scope |
+| Public surface or package boundary identified | yes | `packages/kitcn` cRPC React infinite-query adapter |
+| Convex entry/import graph impact identified | yes | Client-only React/Solid hooks import one type-only/internal helper; no Convex function entry graph |
+| CLI/scaffold/generated impact identified | yes | No CLI/scaffold source changed; required fixture sync refreshed six lucide package snapshots; a raw Next scenario fixture needed a client boundary for current Next |
+| Release artifact path selected | yes | `.changeset/calm-pages-wait.md` |
+| `changeset` skill loaded when `.changeset` is required | yes | Loaded skill and `.agents/rules/changeset.mdc` before changeset creation |
+| Package build / fixture impact decision recorded | yes | `packages/kitcn` build required; fixture sync/check required after current generator drift appeared |
+
+Work Checklist:
+- [x] Timed work N/A: no duration requested; final evidence confidence is 97%.
+- [x] Objective, threshold, verification, constraints, boundaries, and blocker
+      are recorded.
+- [x] Discussion source, acceptance cases, package owner, browser waiver, and
+      root-cause layer are recorded.
+- [x] Video evidence N/A: discussion has no attachment or recording.
+- [x] Public behavior claim challenged and reproduced before implementation.
+- [x] Repro ladder stopped at honest focused React/Solid hook harnesses; browser
+      and screenshots are N/A because no visual/native state is involved.
+- [x] Hard-stop rule satisfied: implementation began only after two red repros.
+- [x] Root instructions, doctrine, local Convex source, hook patterns, migration
+      guide, TDD, changeset, scenarios, and autoreview instructions were read.
+- [x] Three source-listed cases have owner, harness, verdict, and evidence.
+- [x] Readiness classified `ready` from direct source/test evidence.
+- [x] Shared internal split policy is the durable owner; no caller shim.
+- [x] Patch changeset added for published `kitcn` runtime behavior.
+- [x] Bug handoff, task-style PR body, and discussion sync shape are recorded.
+- [x] Implementation committed as `0e6caccd`; branch pushed and PR #305 opened.
+- [x] PR #270 task-style body created for PR #305 and read back with
+      `gh pr view 305 --json body`.
+- [x] Dedicated `codex/fix-native-convex-pagination` branch used.
+- [x] Local-env retry recorded: one `bun install`, exact `bun check` rerun;
+      remaining raw fixture failure was proven on `origin/main` and repaired.
+- [x] All proof ran in `/Users/zbeyens/git/better-convex` or its owning
+      `packages/kitcn` workspace.
+- [x] Searches used exact symbols and bounded output; one broad `rg` truncated,
+      then the investigation switched to exact fixture files.
+- [x] Runtime risk recorded: false splitting can exhaust pagination early;
+      parity predicate plus 3-of-6 and server-recommended cases prove boundary.
+- [x] Final `autoreview --mode local` completed clean with no accepted or
+      actionable findings; overall 0.94.
+- [x] Agent-native review N/A: no agent rules, skills, hooks, commands, or
+      user-action tooling changed.
+- [x] Docs lane and source owner recorded; no docs edit because current guide
+      already demonstrates native pagination through cRPC.
+- [x] Named docs/API claims were checked against source.
+- [x] Docs voice/links/previews N/A: no public content changed.
+- [x] Package boundary and release impact recorded; no public export changed.
+- [x] Release matrix applied with `.changeset/calm-pages-wait.md`.
+- [x] Changeset skill/rule loaded and patch prose follows required format.
+- [x] No-artifact path N/A: published `kitcn` runtime behavior changed.
+- [x] Compatibility decision: preserve native Convex result semantics without
+      changing the hook API or requiring migration.
+- [x] Convex static entry graphs stay unchanged; helper is client-only and its
+      Convex import is type-only.
+- [x] CLI contract N/A: no CLI command changed.
+- [x] Docs/skill sync N/A: public guidance was already correct.
+- [x] Package tests, typecheck, build, lint, fixtures, and full gate are recorded.
+- [x] `packages/kitcn` build and `fixtures:sync`/`fixtures:check` passed.
+
+Completion Gates:
+| Gate | Applies | Required action | Evidence |
+|------|---------|-----------------|----------|
+| Named verification threshold | yes | Run named proof and gates | Focused React 6/6, Solid 5/5, package build/typecheck, fixture sync/check, scenario repair proof, and `bun check` passed |
+| Pre-solution issue challenge verdict | yes | Record challenge | Valid; direct red repros; shared split predicate is durable owner |
+| Repro escalation ladder | yes | Record ladder | Focused hook integration tests were honest and sufficient; browser/visual N/A |
+| Bug reproduced before fix | yes | Record red proof | React and Solid each expected one subscription but received two |
+| Targeted behavior verification | yes | Run focused proof | React 6 pass/29 assertions; Solid 5 pass |
+| TypeScript or typed config changed | yes | Run typecheck | `bun --cwd packages/kitcn typecheck` and root `bun check` typecheck passed |
+| Package exports or file layout changed | yes | Run build | New internal helper; `bun --cwd packages/kitcn build` passed |
+| Package manifests, lockfile, or install graph changed | yes | Install/check | Six generated fixture manifests refreshed; `bun install` and `bun check` passed; root lockfile unchanged |
+| Agent rules or skills changed | no | N/A | No agent-owned source changed; postinstall sync produced no diff |
+| Workspace authority proof | yes | Verify in owner | All commands ran in repo root or `packages/kitcn` |
+| Browser surface changed | no | N/A | Hook has no visual surface |
+| Browser final proof | no | N/A | Automated hook/runtime harness is the honest proof |
+| Scaffold or fixture output changed | yes | Sync/check | `fixtures:sync` refreshed six lucide snapshots; `fixtures:check` and `bun check` passed |
+| Package behavior or public API changed | yes | Add changeset | Patch changeset added for runtime fix |
+| Docs and kitcn skill sync changed | no | N/A | No guidance changed; existing migration guide is correct |
+| Docs or content changed | no | N/A | Only this goal plan changed |
+| High-risk mini gate | yes | Record risk/proof | Risk is premature exhaustion; 3-of-6 plus split-recommended tests cover both sides |
+| Agent-native review for agent/tooling changes | no | N/A | No agent/tooling contract changed |
+| Local install corruption suspected | yes | Retry once | `bun install` rerun did not hide product failure; stale mainline raw fixture then repaired and proved |
+| Commit created | yes | Commit entire checkout | `0e6caccd fix native Convex pagination` |
+| PR create or update | yes | Push and open PR after `bun check` | PR #305 open from `codex/fix-native-convex-pagination` to `main` |
+| Task-style PR body verified | yes | Read body back | `gh pr view 305 --json body` confirmed auto-release block, fix/confidence lines, proof table, and all four emoji sections |
+| PR proof image hosting | no | N/A | No browser image applies |
+| GitHub issue sync-back | yes | Post concise sync after PR | Discussion comment `DC_kwDOPTlS684BDuxx` read back from discussion #304 |
+| Final handoff contract | yes | Fill exact handoff fields | PR #305, discussion comment, 97% confidence, test/browser/outcome/caveat/design evidence recorded below |
+| Final lint | yes | Run lint | `bun lint:fix` and final `bun check` lint passed |
+| Output budget discipline | yes | Record recovery | One broad `rg` truncated; subsequent reads were exact and bounded |
+| Timed checkpoint | no | N/A | No duration requested |
+| Autoreview for non-trivial implementation changes | yes | Run required review | Final local autoreview clean; no accepted/actionable findings; overall 0.94 |
+| Goal plan complete | yes | Run `node .agents/skills/autogoal/scripts/check-complete.mjs docs/plans/304-native-convex-pagination-in-crpc-infinite-queries.md` | passed |
+| Docs source-backed claim audit | yes | Audit existing claims | Migration guide already shows native `paginate()` through cRPC |
+| Docs links / routes / previews | no | N/A | No docs content changed |
+| Docs MDX/content parser | no | N/A | No MDX/content changed |
+| Kitcn docs sync | no | N/A | No `www/**` change |
+| Public API / package boundary proof | yes | Audit boundary | Public exports/signatures unchanged; internal predicate shared by React/Solid |
+| Convex bundle/import proof | yes | Audit graph | No Convex function entry changed; Convex import is type-only |
+| CLI/scaffold/generated proof | yes | Regenerate/check | CLI source unchanged; required fixture sync/check passed |
+| Release artifact classification | yes | Classify delta | Published `kitcn` runtime bug fix |
+| Published package changeset | yes | Add changeset | `.changeset/calm-pages-wait.md` |
+| No release artifact | no | N/A | Published behavior changed |
+| Package typecheck/build/test | yes | Run owning checks | Focused suites, typecheck, build, and full `bun check` passed |
+| Fixture/scaffold generation | yes | Sync/check | Both commands passed; six package snapshots refreshed |
+| Docs/package skill sync | no | N/A | No public guidance changed |
+
+Phase / pass table:
+| Phase | Status | Evidence | Next |
+|-------|--------|----------|------|
+| Intake and source read | complete | Discussion, Convex source, hooks, tests, and migration docs read | source-level repro |
+| Implementation | complete | React and Solid red-green tests; shared split predicate; patch changeset | verification |
+| Verification | complete | Focused tests, package checks, fixtures, repaired raw scenario, and full `bun check` passed | review |
+| Commit / PR / GitHub sync | complete | Commit `0e6caccd`, PR #305, task body read-back, discussion comment read-back | closeout |
+| Closeout | complete | Final contract filled; closeout check follows this edit | final response |
+
+Findings:
+- Discussion #304 has no comments, answer, attachment, or video.
+- The observed case is native Convex pagination through a partially migrated
+  cRPC query; an ORM/schema migration is not a valid prerequisite.
+- Convex returns `splitCursor` as a possible split point, not a command.
+  Convex's clients split only for `SplitRecommended`, `SplitRequired`, or page
+  growth beyond twice the initial item count.
+- Kitcn's React and Solid hooks treated every `splitCursor` as a command,
+  causing the 3 -> 5 -> 6 automatic accumulation.
+- Current fixture generation refreshed lucide-react from `^1.25.0` to
+  `^1.26.0` in six committed snapshots.
+- The full runtime gate exposed a pre-existing raw Next fixture without the
+  client boundary required by current Next; `origin/main` had the same source.
+
+Decisions and tradeoffs:
+- Centralize the predicate under `packages/kitcn/src/internal/pagination.ts`
+  -> React and Solid share Convex parity -> avoids a future framework drift.
+- Do not edit migration docs -> they already state incremental coexistence and
+  show cRPC returning native `ctx.db.query(...).paginate(...)` -> runtime was
+  wrong, not guidance.
+
+Implementation notes:
+- A plain native page with three items, `isDone: false`, `continueCursor`, and
+  `splitCursor` remains one subscription until explicit `fetchNextPage()`.
+- Server-requested and oversized-page splitting remain enabled by the shared
+  Convex predicate.
+
+Review fixes:
+- Added `limit` to the React split effect dependencies after lint caught the
+  missing dependency.
+- Added `'use client'` to the raw Next/shadcn fixture button after the current
+  Next runtime rejected Radix `createContext` from a server component.
+- Final autoreview: clean, no accepted/actionable findings, overall 0.94.
+
+Error attempts:
+| Error / failed attempt | Count | Next different move | Resolution |
+|------------------------|-------|---------------------|------------|
+| First focused React/Solid repro | 2 | Implement shared Convex parity predicate | Both framework suites green |
+| First `fixtures:sync` stalled in external sparse clone | 1 | Stop stalled process and retry once | Retry completed; fixture check passed |
+| First `bun check` found missing React effect dependency and fixture drift | 1 | Fix dependency; run owned fixture sync | Lint and fixtures passed |
+| Runtime scenario resolved mixed/stale React state | 1 | Run the single allowed `bun install`; rerun exact gate | Install completed; exact gate exposed same raw fixture defect |
+| Raw Next fixture returned 500 on current Next | 2 | Compare with `origin/main`, add required client boundary, run focused scenario then full gate | Focused scenario returned 200; full `bun check` passed |
+
+Verification evidence:
+- RED: `bun test packages/kitcn/src/react/use-infinite-query.test.tsx` -> 4
+  pass, 1 fail; expected one query, received two.
+- RED: `bunx vitest run
+  packages/kitcn/src/solid/use-infinite-query.vitest.tsx` -> 4 pass, 1 fail;
+  expected one query, received two.
+- GREEN after refactor: same React command -> 5 pass; same Solid command -> 5
+  pass.
+- FINAL focused React: 6 pass, 29 assertions, including plain split cursor and
+  Convex-recommended split cases; Solid: 5 pass.
+- `bun --cwd packages/kitcn typecheck`, `bun --cwd packages/kitcn build`, and
+  `bun lint:fix` passed.
+- `bun run fixtures:sync` and `bun run fixtures:check` passed; six package
+  snapshots now match current generator output.
+- `bun run scenario:test -- create-convex-nextjs-shadcn` passed with HTTP 200.
+- Final `bun check` passed: lint, root typecheck, Bun/Vitest/CLI/Concave tests,
+  fixtures, verify lane, all runtime and auth smoke scenarios.
+
+Source-listed case matrix:
+| Case | Source claim | Harness | Before | Expected after | Evidence | Status |
+| --- | --- | --- | --- | --- | --- | --- |
+| Page accumulation | Six rows with limit three appear as 3, 5, then 6 because the hook auto-fetches | Focused React infinite-query test with native Convex page results | Query count became two before `fetchNextPage()` | Exactly 3 after first request and 6 only after one explicit next-page request | React focused test: 6 pass, 29 assertions | fixed |
+| Next-page state | `hasNextPage` is always false | Same focused hook test | Plain split cursor forced aggregation toward exhaustion | true after first 3-row page; false after second/exhausted page | React focused test asserts true then false | fixed |
+| Partial migration | Native Convex schema/paginator cannot easily be replaced with ORM | React and Solid native `PaginationResult` hook harnesses plus migration source audit | Both hooks created a split subscription from native result metadata | Native Convex paginator works without ORM/schema conversion | Both framework tests pass; migration guide already uses native paginator | fixed |
+
+Final handoff contract:
+- Commit line: `0e6caccd fix native Convex pagination`
+- PR line: https://github.com/udecode/kitcn/pull/305
+- Issue line:
+  https://github.com/udecode/kitcn/discussions/304#discussioncomment-17755249
+- Confidence line: 🟢 97% — focused regression, package, fixture, runtime, full
+  gate, and clean autoreview evidence
+- Flow table:
+  - Reproduced: 🔴 React/Solid made a second subscription from plain
+    `splitCursor`; browser N/A
+  - Verified: 🟢 React 6/6, Solid 5/5, `bun check`; browser N/A
+- Browser check: N/A: no visual surface; direct hook/runtime harness is honest
+- Outcome: native Convex pagination stays at 3 rows until explicit
+  `fetchNextPage()`, then reaches 6 and flips `hasNextPage` to false
+- Caveat: fixture sync refreshed six lucide ranges; current Next also required
+  a client boundary in the pre-existing raw shadcn runtime fixture
+- Design:
+  - Chosen boundary: shared internal Convex split predicate
+  - Why not quick patch: hook-local conditionals would drift across frameworks
+  - Why not broader change: native pagination contract is sufficient; no ORM,
+    schema, public API, or docs migration is needed
+- Verified: focused tests, package typecheck/build, lint, fixture sync/check,
+  focused runtime scenario, full `bun check`, final autoreview clean at 0.94
+- PR body verified: yes, read back from PR #305
+
+Task-style PR body contract:
+- Preserve any existing `<!-- auto-release:start -->` block. If a changeset is
+  part of the diff and repo policy expects auto release, include that block.
+- Use the accepted PR #270 visual format. The body starts with an emoji
+  issue/fix line, for example `🐛 Fixes #123` or `🐛 Fixes ➖ N/A`, then
+  an emoji confidence line like `🟢 95-100% confidence`.
+- Use this exact table header: `| Phase | 🧪 Tests | 🌐 Browser |`.
+- Use `Reproduced` and `Verified` rows. Mark passing proof with `🟢`, repro or
+  failing proof with `🔴`, and non-applicable cells with `➖ N/A`.
+- Use bold emoji section headings: `**✅ Outcome**`, `**⚠️ Caveat**`,
+  `**🏗️ Design**`, and `**🧪 Verified**`.
+- Never include a line that links to the current PR itself. The current PR URL
+  belongs in the final response, not in its own description.
+- Do not replace this with a generic `Summary` / `Verification` PR body, an
+  adaptive prose body from a git helper skill, plain `## Outcome` sections, or
+  an unrelated generated badge footer unless the caller or repo template
+  explicitly asks for it.
+- Proof is `gh pr view --json body` output or a concise source-backed summary
+  of that output.
+
+Final handoff / sync:
+- Commit: `0e6caccd`
+- PR: https://github.com/udecode/kitcn/pull/305
+- Issue:
+  https://github.com/udecode/kitcn/discussions/304#discussioncomment-17755249
+- Browser proof: N/A: no visual surface
+- Caveats: six generated fixture manifest refreshes and one stale raw Next
+  fixture client-boundary repair shipped with the pagination fix
+
+Timeline:
+- 2026-07-23T17:39:30.782Z Task goal plan created.
+- 2026-07-23T17:44Z React repro failed 1/5 at expected one query,
+  received two; predicate fix passed 5/5.
+- 2026-07-23T17:45Z Solid repro failed 1/5 at expected one query,
+  received two; predicate fix passed 5/5.
+- 2026-07-23T17:45Z Shared predicate refactor passed both focused suites.
+- 2026-07-23T20:22Z Current Next rejected the stale raw fixture server/client
+  boundary; focused scenario passed after the one-line fixture repair.
+- 2026-07-23T20:28Z Final `bun check` passed every repo and runtime gate.
+- 2026-07-23T20:31Z Final autoreview clean at 0.94; commit `0e6caccd`
+  pushed; PR #305 opened and discussion #304 synced/read back.
+- 2026-07-23T20:33Z Goal-plan completion check passed.
+
+Reboot status:
+| Question | Answer |
+|----------|--------|
+| Where am I? | Closeout |
+| Where am I going? | Goal-plan completion check, closeout commit/push, final response |
+| What is the goal? | Fix and ship discussion #304 with direct 3-of-6 pagination proof |
+| What have I learned? | Plain `splitCursor` is metadata; page status/size owns splitting |
+| What have I done? | Reproduced/fixed both hooks, added changeset, restored stale runtime fixture, passed `bun check` |
+
+Open risks:
+- The shared predicate must retain both ordinary and server-recommended split
+  paths; focused tests cover both.
+- Fixture snapshot version refresh is generated collateral, not pagination
+  behavior; `fixtures:check` proves the committed snapshots match the owner.
+
+Hard closeout guard:
+- A local-only final response for verified code-changing work is invalid unless
+  this plan records an explicit user decline, no local patch, analytical/
+  blocked/inconclusive outcome, or a real commit/PR blocker.
diff --git a/fixtures/next-auth/package.json b/fixtures/next-auth/package.json
index f98cd525c..1a74d8a17 100644
--- a/fixtures/next-auth/package.json
+++ b/fixtures/next-auth/package.json
@@ -9,7 +9,7 @@
     "convex": "1.38.0",
     "hono": "4.12.9",
     "kitcn": "workspace:*",
-    "lucide-react": "^1.25.0",
+    "lucide-react": "^1.26.0",
     "next": "16.2.6",
     "next-themes": "^0.4.6",
     "react": "19.2.4",
diff --git a/fixtures/next/package.json b/fixtures/next/package.json
index e64e6b7b9..5a22793d4 100644
--- a/fixtures/next/package.json
+++ b/fixtures/next/package.json
@@ -8,7 +8,7 @@
     "convex": "1.38.0",
     "hono": "4.12.9",
     "kitcn": "workspace:*",
-    "lucide-react": "^1.25.0",
+    "lucide-react": "^1.26.0",
     "next": "16.2.6",
     "next-themes": "^0.4.6",
     "react": "19.2.4",
diff --git a/fixtures/start-auth/package.json b/fixtures/start-auth/package.json
index 3bdcc872c..5347da2ba 100644
--- a/fixtures/start-auth/package.json
+++ b/fixtures/start-auth/package.json
@@ -17,7 +17,7 @@
     "convex": "1.38.0",
     "hono": "4.12.9",
     "kitcn": "workspace:*",
-    "lucide-react": "^1.25.0",
+    "lucide-react": "^1.26.0",
     "react": "^19.2.6",
     "react-dom": "^19.2.6",
     "shadcn": "latest",
diff --git a/fixtures/start/package.json b/fixtures/start/package.json
index e9d553964..235150c0d 100644
--- a/fixtures/start/package.json
+++ b/fixtures/start/package.json
@@ -16,7 +16,7 @@
     "convex": "1.38.0",
     "hono": "4.12.9",
     "kitcn": "workspace:*",
-    "lucide-react": "^1.25.0",
+    "lucide-react": "^1.26.0",
     "react": "^19.2.6",
     "react-dom": "^19.2.6",
     "shadcn": "latest",
diff --git a/fixtures/vite-auth/package.json b/fixtures/vite-auth/package.json
index a22df2be1..516819975 100644
--- a/fixtures/vite-auth/package.json
+++ b/fixtures/vite-auth/package.json
@@ -11,7 +11,7 @@
     "convex": "1.38.0",
     "hono": "4.12.9",
     "kitcn": "workspace:*",
-    "lucide-react": "^1.25.0",
+    "lucide-react": "^1.26.0",
     "react": "^19.2.6",
     "react-dom": "^19.2.6",
     "shadcn": "latest",
diff --git a/fixtures/vite/package.json b/fixtures/vite/package.json
index b9f53bee1..062f31fcb 100644
--- a/fixtures/vite/package.json
+++ b/fixtures/vite/package.json
@@ -10,7 +10,7 @@
     "convex": "1.38.0",
     "hono": "4.12.9",
     "kitcn": "workspace:*",
-    "lucide-react": "^1.25.0",
+    "lucide-react": "^1.26.0",
     "react": "^19.2.6",
     "react-dom": "^19.2.6",
     "shadcn": "latest",
diff --git a/packages/kitcn/src/internal/pagination.ts b/packages/kitcn/src/internal/pagination.ts
new file mode 100644
index 000000000..c942381ae
--- /dev/null
+++ b/packages/kitcn/src/internal/pagination.ts
@@ -0,0 +1,14 @@
+import type { PaginationResult } from 'convex/server';
+
+type SplitPaginationResult<T> = PaginationResult<T> & {
+  splitCursor: string;
+};
+
+export const shouldSplitPaginationPage = <T>(
+  page: PaginationResult<T>,
+  initialNumItems?: number
+): page is SplitPaginationResult<T> =>
+  Boolean(page.splitCursor) &&
+  (page.pageStatus === 'SplitRecommended' ||
+    page.pageStatus === 'SplitRequired' ||
+    (initialNumItems !== undefined && page.page.length > initialNumItems * 2));
diff --git a/packages/kitcn/src/react/use-infinite-query.test.tsx b/packages/kitcn/src/react/use-infinite-query.test.tsx
index 270ab76e1..27f8a54f5 100644
--- a/packages/kitcn/src/react/use-infinite-query.test.tsx
+++ b/packages/kitcn/src/react/use-infinite-query.test.tsx
@@ -174,6 +174,87 @@ describe('useInfiniteQuery', () => {
     expect((firstCall.queries[0] as any).enabled).toBe(true);
   });
 
+  test('does not split a native Convex page solely because it has a split cursor', () => {
+    const queryClient = new QueryClient();
+    const wrapper = makeWrapper(queryClient);
+    const firstPage = {
+      page: [{ _id: 'post-1' }, { _id: 'post-2' }, { _id: 'post-3' }],
+      isDone: false,
+      continueCursor: 'cursor-3',
+      splitCursor: 'cursor-2',
+    };
+    const secondPage = {
+      page: [{ _id: 'post-4' }, { _id: 'post-5' }, { _id: 'post-6' }],
+      isDone: true,
+      continueCursor: 'cursor-6',
+    };
+
+    useQueriesSpy.mockImplementation((arg: UseQueriesArg) => {
+      useQueriesCalls.push(arg);
+      return (arg as any).combine(
+        [firstPage, secondPage]
+          .slice(0, arg.queries.length)
+          .map((data, index) => ({
+            data,
+            dataUpdatedAt: index + 1,
+            isError: false,
+            isFetching: false,
+            isLoading: false,
+            isPlaceholderData: false,
+          }))
+      );
+    });
+
+    const options = createOptions({ limit: 3 });
+    const { result } = renderHook(() => useInfiniteQuery(options), { wrapper });
+
+    expect(result.current.data).toEqual(firstPage.page);
+    expect(result.current.hasNextPage).toBe(true);
+    expect(useQueriesCalls.at(-1)?.queries).toHaveLength(1);
+
+    act(() => {
+      result.current.fetchNextPage();
+    });
+
+    expect(result.current.data).toEqual([
+      ...firstPage.page,
+      ...secondPage.page,
+    ]);
+    expect(result.current.hasNextPage).toBe(false);
+    expect(useQueriesCalls.at(-1)?.queries).toHaveLength(2);
+  });
+
+  test('splits a page when Convex recommends it', () => {
+    const queryClient = new QueryClient();
+    const wrapper = makeWrapper(queryClient);
+    const firstPage = {
+      page: [{ _id: 'post-1' }, { _id: 'post-2' }, { _id: 'post-3' }],
+      isDone: false,
+      continueCursor: 'cursor-3',
+      splitCursor: 'cursor-2',
+      pageStatus: 'SplitRecommended',
+    };
+
+    useQueriesSpy.mockImplementation((arg: UseQueriesArg) => {
+      useQueriesCalls.push(arg);
+      return (arg as any).combine([
+        {
+          data: firstPage,
+          dataUpdatedAt: 1,
+          isError: false,
+          isFetching: false,
+          isLoading: false,
+          isPlaceholderData: false,
+        },
+      ]);
+    });
+
+    const options = createOptions({ limit: 3 });
+    renderHook(() => useInfiniteQuery(options), { wrapper });
+
+    expect(useQueriesCalls.at(-1)?.queries).toHaveLength(2);
+  });
+
   test('fetchNextPage adds a new page query with continueCursor and limit', () => {
     const queryClient = new QueryClient();
     const wrapper = makeWrapper(queryClient);
diff --git a/packages/kitcn/src/react/use-infinite-query.ts b/packages/kitcn/src/react/use-infinite-query.ts
index b9059e0d9..0744cadda 100644
--- a/packages/kitcn/src/react/use-infinite-query.ts
+++ b/packages/kitcn/src/react/use-infinite-query.ts
@@ -23,6 +23,7 @@ import { useCallback, useEffect, useMemo, useRef, useState } from 'react';
 import { CRPCClientError, isCRPCClientError } from '../crpc/error';
 import { convexQuery } from '../crpc/query-options';
 import { type ExtractPaginatedItem, FUNC_REF_SYMBOL } from '../crpc/types';
+import { shouldSplitPaginationPage } from '../internal/pagination';
 import type { DistributiveOmit } from '../internal/types';
 import { useAuthValue, useSafeConvexAuth } from './auth-store';
 import { useMeta } from './context';
@@ -134,13 +135,6 @@ type PageState = {
   endCursor?: string | null; // For page splitting - the cursor where this page ends
 };
 
-// Page splitting: when a page gets too large, Convex may return splitCursor
-// - SplitRecommended: page is large, should split on next render
-// - SplitRequired: page MUST be split (too large to return)
-type PageResultWithSplit<T> = PaginationResult<T> & {
-  splitCursor?: string | null;
-};
-
 /** Build a unique key for recovery attempt detection */
 const buildRecoveryKey = (
   pageKeys: number[],
@@ -507,7 +501,7 @@ const useInfiniteQueryInternal = <Query extends PaginatedQueryReference>(
       const allItems: PaginatedQueryItem<Query>[] = [];
       const pages: PaginatedQueryItem<Query>[][] = [];
       const seenIds = new Set<string>();
-      let lastPage: PageResultWithSplit<PaginatedQueryItem<Query>> | undefined;
+      let lastPage: PaginationResult<PaginatedQueryItem<Query>> | undefined;
       let paginationStatus: PaginationStatus = 'LoadingFirstPage';
 
       for (let i = 0; i < results.length; i++) {
@@ -516,7 +510,7 @@ const useInfiniteQueryInternal = <Query extends PaginatedQueryReference>(
           paginationStatus = i === 0 ? 'LoadingFirstPage' : 'LoadingMore';
           break;
         }
-        const page = pageQuery.data as PageResultWithSplit<
+        const page = pageQuery.data as PaginationResult<
           PaginatedQueryItem<Query>
         >;
         lastPage = page;
@@ -577,25 +571,29 @@ const useInfiniteQueryInternal = <Query extends PaginatedQueryReference>(
     state,
   });
 
-  // Handle page splitting - when a page returns splitCursor, we need to split it
+  // Split when Convex requests it or a reactive page outgrows its target size.
   useEffect(() => {
     for (let i = 0; i < combined._rawResults.length; i++) {
       const pageQuery = combined._rawResults[i];
       if (pageQuery.data) {
-        const page = pageQuery.data as PageResultWithSplit<
+        const page = pageQuery.data as PaginationResult<
           PaginatedQueryItem<Query>
         >;
         const pageKey = state.pageKeys[i];
         const pageState = state.queries[pageKey];
 
         // Check if this page needs splitting and we haven't already split it
-        if (page.splitCursor && pageState && !pageState.endCursor) {
+        if (
+          shouldSplitPaginationPage(page, limit) &&
+          pageState &&
+          !pageState.endCursor
+        ) {
           setState((prev) => {
             const currentPageState = prev.queries[pageKey];
             if (!currentPageState || currentPageState.endCursor) return prev;
 
             const newKey = prev.nextPageKey;
-            const splitCursor = page.splitCursor!; // Checked above: page.splitCursor is truthy
+            const splitCursor = page.splitCursor;
             const splitPageArgs = {
               ...argsObject,
               cursor: splitCursor,
@@ -635,6 +633,7 @@ const useInfiniteQueryInternal = <Query extends PaginatedQueryReference>(
     state.pageKeys,
     state.queries,
     argsObject,
+    limit,
     setState,
   ]);
 
diff --git a/packages/kitcn/src/solid/use-infinite-query.ts b/packages/kitcn/src/solid/use-infinite-query.ts
index 060b3d599..f832b34aa 100644
--- a/packages/kitcn/src/solid/use-infinite-query.ts
+++ b/packages/kitcn/src/solid/use-infinite-query.ts
@@ -20,6 +20,7 @@ import { createEffect, createMemo, createSignal, on } from 'solid-js';
 import { CRPCClientError, isCRPCClientError } from '../crpc/error';
 import { convexQuery } from '../crpc/query-options';
 import { type ExtractPaginatedItem, FUNC_REF_SYMBOL } from '../crpc/types';
+import { shouldSplitPaginationPage } from '../internal/pagination';
 import type { DistributiveOmit } from '../internal/types';
 import { useMeta } from './auth';
 import { useAuthValue, useSafeConvexAuth } from './auth-store';
@@ -139,13 +140,6 @@ type PageState = {
   endCursor?: string | null; // For page splitting - the cursor where this page ends
 };
 
-// Page splitting: when a page gets too large, Convex may return splitCursor
-// - SplitRecommended: page is large, should split on next render
-// - SplitRequired: page MUST be split (too large to return)
-type PageResultWithSplit<T> = PaginationResult<T> & {
-  splitCursor?: string | null;
-};
-
 /** Build a unique key for recovery attempt detection */
 const buildRecoveryKey = (
   pageKeys: number[],
@@ -500,7 +494,7 @@ const useInfiniteQueryInternal = <Query extends PaginatedQueryReference>(
       const allItems: PaginatedQueryItem<Query>[] = [];
       const pages: PaginatedQueryItem<Query>[][] = [];
       const seenIds = new Set<string>();
-      let lastPage: PageResultWithSplit<PaginatedQueryItem<Query>> | undefined;
+      let lastPage: PaginationResult<PaginatedQueryItem<Query>> | undefined;
       let paginationStatus: PaginationStatus = 'LoadingFirstPage';
 
       for (let i = 0; i < results.length; i++) {
@@ -509,7 +503,7 @@ const useInfiniteQueryInternal = <Query extends PaginatedQueryReference>(
           paginationStatus = i === 0 ? 'LoadingFirstPage' : 'LoadingMore';
           break;
         }
-        const page = pageQuery.data as PageResultWithSplit<
+        const page = pageQuery.data as PaginationResult<
           PaginatedQueryItem<Query>
         >;
         lastPage = page;
@@ -558,7 +552,7 @@ const useInfiniteQueryInternal = <Query extends PaginatedQueryReference>(
   })) as {
     data: PaginatedQueryItem<Query>[];
     dataUpdatedAt: number;
-    lastPage: PageResultWithSplit<PaginatedQueryItem<Query>> | undefined;
+    lastPage: PaginationResult<PaginatedQueryItem<Query>> | undefined;
     pages: PaginatedQueryItem<Query>[][];
     status: PaginationStatus;
     error: Error | null;
@@ -581,7 +575,7 @@ const useInfiniteQueryInternal = <Query extends PaginatedQueryReference>(
     state,
   });
 
-  // Handle page splitting - when a page returns splitCursor, we need to split it
+  // Split when Convex requests it or a reactive page outgrows its target size.
   createEffect(
     on(
       [
@@ -594,21 +588,25 @@ const useInfiniteQueryInternal = <Query extends PaginatedQueryReference>(
         for (let i = 0; i < combined._rawResults.length; i++) {
           const pageQuery = combined._rawResults[i];
           if (pageQuery.data) {
-            const page = pageQuery.data as PageResultWithSplit<
+            const page = pageQuery.data as PaginationResult<
               PaginatedQueryItem<Query>
             >;
             const pageKey = state().pageKeys[i];
             const pageState = state().queries[pageKey];
 
             // Check if this page needs splitting and we haven't already split it
-            if (page.splitCursor && pageState && !pageState.endCursor) {
+            if (
+              shouldSplitPaginationPage(page, limit) &&
+              pageState &&
+              !pageState.endCursor
+            ) {
               setState((prev) => {
                 const currentPageState = prev.queries[pageKey];
                 if (!currentPageState || currentPageState.endCursor)
                   return prev;
 
                 const newKey = prev.nextPageKey;
-                const splitCursor = page.splitCursor!;
+                const splitCursor = page.splitCursor;
                 const splitPageArgs = {
                   ...argsObject(),
                   cursor: splitCursor,
diff --git a/packages/kitcn/src/solid/use-infinite-query.vitest.tsx b/packages/kitcn/src/solid/use-infinite-query.vitest.tsx
index 127df8556..372fa0dee 100644
--- a/packages/kitcn/src/solid/use-infinite-query.vitest.tsx
+++ b/packages/kitcn/src/solid/use-infinite-query.vitest.tsx
@@ -201,6 +201,41 @@ describe('useInfiniteQuery', () => {
     expect((firstCall.queries[0] as any).enabled).toBe(true);
   });
 
+  test('does not split a native Convex page solely because it has a split cursor', () => {
+    const queryClient = new QueryClient();
+    const wrapper = makeWrapper(queryClient);
+    const firstPage = {
+      page: [{ _id: 'post-1' }, { _id: 'post-2' }, { _id: 'post-3' }],
+      isDone: false,
+      continueCursor: 'cursor-3',
+      splitCursor: 'cursor-2',
+    };
+    let capturedAccessor: (() => UseQueriesArg) | null = null;
+
+    mockUseQueries.mockImplementation((accessor: any) => {
+      capturedAccessor = accessor;
+      const arg = typeof accessor === 'function' ? accessor() : accessor;
+      useQueriesCalls.push(arg);
+      return arg.combine([
+        {
+          data: firstPage,
+          dataUpdatedAt: 1,
+          isError: false,
+          isFetching: false,
+          isLoading: false,
+          isPlaceholderData: false,
+        },
+      ]);
+    });
+
+    const options = createOptions({ limit: 3 });
+    const { result } = renderHook(() => useInfiniteQuery(options), { wrapper });
+
+    expect(result.data).toEqual(firstPage.page);
+    expect(result.hasNextPage).toBe(true);
+    expect(capturedAccessor!().queries).toHaveLength(1);
+  });
+
   test('fetchNextPage adds a new page query with continueCursor and limit', () => {
     const queryClient = new QueryClient();
     const wrapper = makeWrapper(queryClient);
diff --git a/tooling/scenario-fixtures/create-convex-nextjs-shadcn/components/ui/button.tsx b/tooling/scenario-fixtures/create-convex-nextjs-shadcn/components/ui/button.tsx
index 04f5fdb4f..486e29de5 100644
--- a/tooling/scenario-fixtures/create-convex-nextjs-shadcn/components/ui/button.tsx
+++ b/tooling/scenario-fixtures/create-convex-nextjs-shadcn/components/ui/button.tsx
@@ -1,3 +1,5 @@
+'use client';
+
 import { Slot } from '@radix-ui/react-slot';
 import { cva, type VariantProps } from 'class-variance-authority';
 import * as React from 'react';