Take over the three stuck dependabot PRs (#1963 tsx, #1962 fumadocs-core, #1058 better-sqlite3): bring each onto main and land it, or close it with the reason

[objectstack-fleet]

Filing gate: ③ a task the maintainer ordered directly. The maintainer, in the repo:hotcrm seat's chat (session_01ER8ntXZhYebyQ66aXWdjfT) at 2026-10-03T03:05Z, verbatim: 「hotcrm一大堆pr你全部接手处理」.

Who acts on it: the repo:hotcrm seat claims this card and dispatches one dev for all three PRs, then lands or closes each. It exists because a claim and a dispatch need a card, and dependabot PRs have none.

The three PRs, read by the seat at 2026-10-03T03:23Z

PR	bump	head	base	CI on head	blocker
#1963	tsx 4.23.12 → 4.23.15 (dev)	d81da1f2	2f7b2326	all green (2026-09-28)	stale base. Dependabot's own rebase failed twice: 「Dependabot tried to update this pull request, but something went wrong」 (5905405385, 5947627898).
#1962	fumadocs-core 16.9.3 → 16.15.14 (dev)	d36259fb	2f7b2326	Build and Test (22.x) failure (job 108748640189); the job log is unreadable from the seat's container	stale base (same dependabot failure, 5905417721, 5947628628), plus a real build failure that has not been diagnosed
#1058	better-sqlite3 12.11.1 → 13.0.3 (major, prod)	7144bf6b	f01e6f6f	all green (2026-09-03)	stale base (dependabot failure 5384490200, 5437018455). os-zhuang measured the bump in 5570617020: no breaking change applies, and the installed @objectstack/* packages already ask for ^13.0.3.

Done when

Each PR is either merged through the merge queue with CI green on a head that contains main, or closed with a one-line reason on the PR (for example, an upstream incompatibility named with its error). Pushes go onto the dependabot branches as merge commits, never force-pushed. Once a human pushes to a branch, dependabot stops managing it, which is accepted here.

Duplicate check

All hotcrm issues, open and closed: 746 issues over 19 REST pages, read to the short page. Title and body were grepped for dependabot|better-sqlite3|fumadocs|\btsx\b: 19 hits, all closed. None is about these three PRs (they are about .github/dependabot.yml labels, @changesets/cli 3.x, docs anchors and similar). Positive control: crm_opportunity hit 106.

---

Generated by Claude Code

[objectstack-fleet]

Claim: PM loop round R71
Session: session_01ER8ntXZhYebyQ66aXWdjfT
Account: hotlong (the seat's linked user as GET /user answers it; always the card's assignee)
Branch: the three dependabot branches, continued with merge commits (⛔ no force-push): dependabot/npm_and_yarn/tsx-4.23.15 (#1963), dependabot/npm_and_yarn/fumadocs-core-16.15.14 (#1962), dependabot/npm_and_yarn/better-sqlite3-13.0.3 (#1058)
Worktree: hotcrm-issue-1993
Domain: repo:hotcrm (single-lane repo, no domain:* taxonomy)
Seat: repo:hotcrm#1
File surface: package.json and pnpm-lock.yaml on each branch, regenerated with pnpm install (⛔ never by hand). For #1962 only, the smallest source fix its build failure needs, if that fix lies in this repo's docs-app code that the fumadocs-core bump breaks. Otherwise, stop and report (stop on breach; explain in the report)
Container & model: M, mode:subagent, model: default (dispatch-gates --tier --repo objectstack-ai/hotcrm answers "no path-derived mandate … the tier stays the PM's per-card judgment call"; a build-failure diagnosis)
Clause-②: no. Dev-dependency and driver-binary bumps touch no published schema and no accept set.
Thread-read: none
Serial constraints cleared: in flight are PR #1953 (the ratchet script and tests), PR #1950 (src/sales/**, docs/STATUS.md, objectstack.composition.ts) and #1369 (src/sales/profiles/**, a test). None touches package.json or pnpm-lock.yaml. Each of the three branches regenerates the lockfile from its own merge with main, and they land one at a time, each merging main again after the previous one lands.

---

Generated by Claude Code

[objectstack-fleet]

os-dev-report

{
"issue": 1993,
"status": "done",
"branch": "dependabot/npm_and_yarn/tsx-4.23.15 | dependabot/npm_and_yarn/fumadocs-core-16.15.14 | dependabot/npm_and_yarn/better-sqlite3-13.0.3",
"pr": null,
"prs": [
"https://github.com//pull/1963",
"https://github.com//pull/1962",
"https://github.com//pull/1058"
],
"session": "session_01ER8ntXZhYebyQ66aXWdjfT (subagent, the parent seat session)",
"premise_still_valid": true,
"summary": "All three dependabot branches now carry a merge commit with main 25cd8d7, pushed straight onto the dependabot branches (the git proxy accepted every push, so no fallback branch and no new PR), and each merged head passes pnpm verify under the shared lock. Hypothesis 1 held on all three: the only conflict was pnpm-lock.yaml, resolved by taking main's side and regenerating with pnpm install (pnpm 10.33.0); every lockfile diff touches only the bumped package's own tree. #1962 was red on the repo's own guard pins fumadocs-core to the version apps/docs renders with (root moved to 16.15.14, apps/docs still 16.9.3); it is fixed by moving apps/docs fumadocs-core and fumadocs-ui to 16.15.14 in lockstep and re-keying the docs-site UI translations for the fumadocs-ui 16.15 i18n API (73 changed source lines). better-sqlite3 13.0.3 is proven to load, not only install: boot on SqlDriver(better-sqlite3), REST create 201 and read 200 on port 4815, and the server process mapped node_modules/better-sqlite3/prebuilds/linux-x64.node.",
"per_pr": [
{
"pr": "#1963 tsx 4.23.12 -> 4.23.15",
"merged_head": "91edbfd2 (merge of d81da1f + 25cd8d7)",
"lockfile_diffstat_vs_main": "pnpm-lock.yaml | 39 ++++++++++++++------------------------- (1 file changed, 14 insertions(+), 25 deletions(-))",
"lockfile_unrelated_moved": "no: tsx 4.23.12 entry removed (4.23.15 was already in main's tree), plus the tsx peer suffix on the vitest / vite / @vitest/mocker snapshot keys",
"verdict_line": "os-verify-lock: VERDICT command-exit 0 · held the lock 177s (2m57s) · waited 0s (same run printed 91edbfd)",
"verdict": "green",
"ci_on_head_one_read_not_waited": "Build and Test (22.x), Quality Checks, Playwright, CodeQL, Analyze Code, Label Pull Request, Vercel Preview Comments: success; Check Changeset: skipped (skip-changeset)"
},
{
"pr": "#1058 better-sqlite3 12.11.1 -> 13.0.3",
"merged_head": "eb561d19 (merge of 7144bf6 + 25cd8d7)",
"lockfile_diffstat_vs_main": "pnpm-lock.yaml | 415 ++------- (1 file changed, 12 insertions(+), 403 deletions(-))",
"lockfile_unrelated_moved": "no: better-sqlite3 12.11.1 and its 12.x-only prebuild-install chain drop out (bindings, chownr, decompress-response, deep-extend, expand-template, file-uri-to-path, github-from-package, ini, mimic-response, mkdirp-classic, napi-build-utils, node-abi, prebuild-install, pump, rc, simple-concat, simple-get, strip-json-comments, tar-fs, tunnel-agent); the better-sqlite3 peer-suffix variants of knex, better-auth, @better-auth/oauth-provider, scim, sso, @objectstack/driver-sqlite-wasm, plugin-auth, runtime, service-datasource and verify collapse onto 13.0.3",
"objectstack_17_6_0_still_asks_for_13": "@objectstack/cli@17.6.0 optionalDependencies.better-sqlite3: ^13.0.3 · @objectstack/driver-sql@17.6.0 optionalDependencies.better-sqlite3: ^13.0.3 ; devDependencies.better-sqlite3: ^13.0.3 · @objectstack/plugin-sharing@17.6.0 devDependencies.better-sqlite3: ^13.0.3 (read from the installed package.json files)",
"runtime_proof": "objectstack start -p 4815 --artifact ./dist/objectstack.json (the artifact this tree built; log: Loading metadata from local artifact file .../hotcrm-issue-1993/dist/objectstack.json, totalRegistered 365) on a fresh scratch DB with --database-driver sqlite. Banner: Driver: SqlDriver(better-sqlite3); zero lines matching step down / native better-sqlite3 unavailable / NODE_MODULE_VERSION. Signed up a probe user, POST /api/v1/data/crm_account -> HTTP 201 (id xyMIfd3WZhSMS3ga, account_number ACC-000010), GET /api/v1/data/crm_account/xyMIfd3WZhSMS3ga -> HTTP 200; the row read back from the sqlite file with the 13.0.3 module (sqlite_version 3.53.4). /proc/PID/maps of the objectstack serve process listed node_modules/better-sqlite3/prebuilds/linux-x64.node. Server PIDs stopped, scratch DB deleted.",
"verdict_line": "os-verify-lock: VERDICT command-exit 0 · held the lock 175s (2m55s) · waited 259s (4m19s) (same run printed eb561d1)",
"verdict": "green",
"ci_on_head_one_read_not_waited": "all checks success; Check Changeset skipped (skip-changeset)"
},
{
"pr": "#1962 fumadocs-core 16.9.3 -> 16.15.14",
"merged_head": "76063f55 (994506b merge of d36259f + 25cd8d7, then c734ae3 apps/docs lockstep + translation re-key, then 76063f5 comment re-scope)",
"lockfile_diffstat_vs_main": "pnpm-lock.yaml | 52 (30 insertions, 22 deletions); apps/docs/pnpm-lock.yaml | 247 (152 insertions, 95 deletions)",
"lockfile_unrelated_moved": "no. Root: fumadocs-core and its own deps only (+@fumari/image-size, npm-to-yarn, zbsearch; -@orama/orama, js-yaml). apps/docs: fumadocs-core / fumadocs-ui 16.15.14 and their own deps only (shiki 4.3.1 -> 4.5.0, motion and framer-motion 12 -> 13, @fumadocs/tailwind 0.0.5 -> 0.1.2, +@fuma-translate/react, cn, zbsearch, npm-to-yarn, yaml, @fumari/image-size, and a nested lucide-react 1.50.0 for fumadocs-ui's own ^1.47.0 dependency; the app's own lucide-react stays 1.27.0); fumadocs-mdx 15.0.10 is re-suffixed only",
"lockstep_basis": "fumadocs-ui@16.15.14 peerDependencies fumadocs-core: 16.15.14 (exact; 16.9.3 pinned 16.9.3), so apps/docs cannot follow the root pin without fumadocs-ui. fumadocs-mdx@15.0.10 peer fumadocs-core ^16.7.0 already admits 16.15.14, so it stays. Lockfile regenerated with pnpm add --save-exact inside apps/docs.",
"docs_app_build": "apps/docs pnpm run types:check && pnpm run build on c734ae3: os-verify-lock: VERDICT command-exit 0 · held the lock 48s · waited 0s; Compiled successfully; Generating static pages 359/359. Before the re-key the same command was VERDICT command-exit 2 on the TS2353 error below.",
"reader_check": "prerendered zh-Hans/docs/administration/automation.html carries 本页目录, 搜索文档, 下一页, 上一页, 选择语言; zh-Hant carries 本頁目錄, 搜尋文檔, 下一頁, 上一頁, 選擇語言 (the English strings left on those pages are the dictionary keys in the RSC payload and the Open Search aria-label, which the app never translated on 16.9.3 either). next start + GET /api/search: en opportunity -> 200, 228 results; zh-Hans 商机 -> 200, 297; zh-Hant 商機 -> 200, 296; zh-Hans 客户 -> 200, 377 (zbsearch now backs search; the @orama/tokenizers mandarin tokenizer still segments).",
"verdict_line": "os-verify-lock: VERDICT command-exit 0 · held the lock 226s (3m46s) · waited 0s (same run printed 76063f5)",
"verdict": "green (lockstep sibling bump + docs-app source fix)",
"ci_on_head_one_read_not_waited": "all checks success, including the docs-app workflow job Typecheck and Build; Check Changeset skipped (skip-changeset)"
}
],
"root_cause_1962": "Reproduced on the merged head 994506b under the lock (VERDICT command-exit 1; Tests 1 failed | 17 passed (18)): FAIL test/docs-anchor-links.test.ts > content/docs anchored links > pins fumadocs-core to the version apps/docs renders with / AssertionError: expected '16.15.14' to be '16.9.3' // Object.is equality at test/docs-anchor-links.test.ts:133:18 expect(here).toBe(site);. Dependabot bumped only the root devDependency, while the guard requires it to equal apps/docs/package.json. The slug rule itself did not move: dist/mdx-plugins/remark-heading.js is byte-identical between fumadocs-core 16.9.3 and 16.15.14, and the other 17 anchor tests pass. Hypothesis 3 was half right: a version skew, but in the repo's own pin guard rather than a sibling build break. The sibling move then exposed a real docs-app break: app/[lang]/layout.tsx(6,3): error TS2353: Object literal may only specify known properties, and 'translations' does not exist in type Partial(Record(\"en\" | \"zh-Hans\" | \"zh-Hant\", Record(string, string) & { displayName?: string })) [angle brackets written as parentheses]. fumadocs-ui 16.15 takes the locale map as the second argument of defineI18nUI and keys each UI string by its English text plus its component (Search(search dialog)) instead of a camelCase id, and ignores unknown keys silently. So the fix re-keys the 12 labels one for one (search maps to both Search(search trigger) and Search(search dialog); chooseLanguage also to its aria-label key), drops chooseTheme (no 16.9.3 component read it), and adds satisfies Partial(Translations) to each locale block so a misspelt or old key is a type error. Reverse check: renaming one key back to editOnGithub turned tsc red with TS2353 on editOnGithub, then restore to HEAD blob 5b2c9230 was proven (git diff HEAD empty).",
"tests": "All gate runs used OS_VERIFY_LOCK_SLOT=hotcrm-issue-1993 bash /home/user/objectstack/scripts/pm/os-verify-lock.sh -c 'git rev-parse --short HEAD && pnpm verify', one per branch, serially. #1963 sha 91edbfd: Validation passed; i18n lint gate 0 i18n/missing-*; source hygiene clean; source token ratchet clean; Build complete; Test Files 174 passed (174), Tests 3701 passed | 1 skipped (3702); VERDICT command-exit 0. #1058 sha eb561d1: the same lines, Test Files 174 passed (174), Tests 3701 passed | 1 skipped (3702); VERDICT command-exit 0. #1962 sha c734ae3: VERDICT command-exit 0 (174/174, 3701 passed), then re-run on the final head after the comment commit, sha 76063f5: Test Files 174 passed (174), Tests 3701 passed | 1 skipped (3702); VERDICT command-exit 0. Targeted: vitest test/docs-anchor-links.test.ts red on 994506b (1 failed | 17 passed) and green on c734ae3 (18 passed, VERDICT command-exit 0). Docs app (CI docs-app.yml shape) types:check + build green on c734ae3. Reverse check and runtime boot as in per_pr. NOT MEASURED locally: pnpm test:coverage (the CI step; CI ran it on all three heads, green).",
"token_ratchet": "before (merged tree, before any lockfile regen) and after (final fumadocs head) are byte-identical logs. src/sales rows: business semantics 45 5,417 213,732 53,433 55,000 · interaction layer 25 3,350 110,585 27,646 31,000 · other authored metadata 45 1,295 61,915 15,479 — · authored total 115 10,062 386,232 96,558 100,000. No ceiling touched, no src/ change.",
"files_changed": {
"#1963": [
"package.json (1 line, from the dependabot commit)",
"pnpm-lock.yaml (regenerated)"
],
"#1058": [
"package.json (1 line, from the dependabot commit)",
"pnpm-lock.yaml (regenerated)"
],
"#1962": [
"package.json (1 line, from the dependabot commit)",
"pnpm-lock.yaml (regenerated)",
"apps/docs/package.json (fumadocs-core, fumadocs-ui)",
"apps/docs/pnpm-lock.yaml (regenerated by pnpm add)",
"apps/docs/app/[lang]/layout.tsx (+38/-34)",
"test/helpers/heading-label.ts (+1, comment)"
]
},
"changeset": "none added. All three PRs already carry skip-changeset; #1962 now also touches the private @hotcrm/docs app and a test-helper comment, nothing in the shipped artifact, so skip-changeset still applies. No governed path (AGENTS.md, CLAUDE.md, .claude, .github/instructions) is in any diff.",
"deviations": [
"Claim file surface and order scope: #1962 carries one comment line in test/helpers/heading-label.ts, outside the declared surface (package.json, pnpm-lock.yaml, docs-app code). The lockstep move made its claim false ("copied verbatim from fumadocs-core@16.9.3 ... the version apps/docs/package.json pins"), so it is re-scoped per AGENTS.md Platform Upgrades rule 5 (re-measured byte-identical in 16.15.14, not renumbered). All four in-place conditions hold: same skew class, mechanical, the file is held by none of #1953 / #1950 / #1369, no new verification surface. Conflict named, not chosen silently: the order said source changes only in docs-app code, the dev standard says fix what your change makes false. The claim surface needs the seat's same-round amendment. It is commit 76063f5 alone; dropping it would take a revert commit (no force-push).",
"Write probe: the branches already existed, so no empty-branch probe push was possible; the first real push (the tsx merge commit) was the probe and was accepted.",
"The tsx merge commit message was amended once before its first push (to drop a card reference from the commit body); no pushed commit was amended, rebased or force-pushed.",
"First reverse-check attempt was a no-op: ablation-replace refused because the replacement contained the anchor, and tsc never ran. It was re-run with a different anchor, which landed and went red.",
"Long gate runs were started detached (nohup) and waited on in the foreground with tail --pid, so the ~10-minute foreground cap could not kill pnpm verify mid-run; the slot was set before the first attempt."
],
"mcp_calls": "0",
"api_writes": "1: POST /repos//issues/1993/comments (this os-dev-report, via post-stamped.mjs). git pushes (not REST): 5 to the three dependabot branches.",
"open_questions": [
{
"question": "#1962 brings fumadocs-ui 16.9.3 -> 16.15.14 into the docs site, so its upstream UI changes ship too (release notes: collapsed-sidebar pill no longer covers the title, search input exposed as a combobox, footer descriptions wrap instead of truncating, copy buttons announce Copied, mark-tag highlighting in search results). Is that past the order's STOP line of a docs-site behaviour change a reader would notice?",
"options": [
"A: land #1962 as pushed. The ruling fixed fumadocs-core at 16.15.14, the repo's guard makes apps/docs follow, and fumadocs-ui pins it exactly; the source fix keeps every localized label (measured in prerendered HTML) and zh search works.",
"B: treat the upstream UI deltas as the STOP line and close #1962 with the reason. The two green PRs are unaffected."
],
"recommendation": "A, because the visible deltas are upstream refinements forced by the bump target the ruling fixed, not by the fix. The fix itself keeps what a reader sees: Chinese chrome, Chinese search, 359 pages building."
}
],
"out_of_scope_findings": [
"carrier: none (承接者:无) · noted, not filed: AGENTS.md (governed) says a NODE_MODULE_VERSION flood on boot is better-sqlite3 built for another Node ABI, fixed by pnpm rebuild. Once #1058 lands, the process loads the N-API prebuild (measured: prebuilds/linux-x64.node mapped), so that advice stops being the likely diagnosis. os-zhuang already noted it on #1058.",
"carrier: none (承接者:无) · noted: pnpm 10.33 still runs node-gyp rebuild for better-sqlite3 13.0.3 on install (binding.gyp ships, the package is in onlyBuiltDependencies, gypfile: false is not honoured). It compiled nothing (stamp TOUCH only) and the prebuild loaded, but the install host runs python3 and make. NOT MEASURED on a host without them.",
"carrier: none (承接者:无) · noted: better-auth 1.7.3 declares an optional peer better-sqlite3 ^12.0.0, unmet by 13.0.3. pnpm warns on every install, and main's tree already warns through @objectstack/cli's nested 13.0.3. No runtime exposure measured: auth sign-up and the session cookie worked on the 13.0.3 SQL boot."
],
"landing_note": "Each branch contains main 25cd8d7 and shows mergeable_state clean. When one lands, the other two need main merged again. tsx and better-sqlite3 both rewrite root lockfile snapshot keys, so expect another pnpm-lock.yaml conflict; the same recipe applies (take main's lockfile, pnpm install). Draft state, auto-merge, labels and assignees on the three PRs were not touched."
}

---

Generated by Claude Code

[objectstack-fleet]

Claim amendment (file surface), repo:hotcrm seat, session_01ER8ntXZhYebyQ66aXWdjfT, 2026-10-03T04:11Z. This amends 5965061112. Every other field stands.

• The chore(deps-dev): bump fumadocs-core from 16.9.3 to 16.15.14 #1962 surface gains one line: a comment re-scope in test/helpers/heading-label.ts (commit 76063f55). The lockstep move to fumadocs-core 16.15.14 made the comment's version statement false. The slug rule it describes was re-measured as byte-identical in 16.15.14. No other PR in flight holds that file.
• What the seat recorded in its read-only check of the three heads:
  • main 25cd8d78 is an ancestor of 91edbfd2, 76063f55 and eb561d19.
  • Every check run is green or skipped, and mergeable_state reads clean on all three.
  • Against main, the diffs are package.json plus pnpm-lock.yaml for chore(deps-dev): bump tsx from 4.23.12 to 4.23.15 #1963 and chore(deps): bump better-sqlite3 from 12.11.1 to 13.0.3 #1058. For chore(deps-dev): bump fumadocs-core from 16.9.3 to 16.15.14 #1962 they are those plus apps/docs/{package.json,pnpm-lock.yaml}, apps/docs/app/[lang]/layout.tsx (the i18n re-key) and that one comment line.
• Open for the maintainer, from the dev's report 5965389725: chore(deps-dev): bump fumadocs-core from 16.9.3 to 16.15.14 #1962 also brings fumadocs-ui 16.15.14's upstream UI refinements into the docs site.
• ⛔ This is not an ACCEPT, and the seat lands nothing here: landing these PRs is with the maintainer.

---

Generated by Claude Code

[objectstack-fleet]

ACCEPT: the three dependabot PRs. #1963 at 91edbfd2, #1058 at eb561d19, #1962 at 76063f55. repo:hotcrm seat, session_01ER8ntXZhYebyQ66aXWdjfT, 2026-10-03T04:54Z. Report: the dev's os-dev-report 5965389725.

Probes taken by the seat itself:

• Ancestry: main 25cd8d78 is an ancestor of all three heads.
• Diffs against main:
  • chore(deps-dev): bump tsx from 4.23.12 to 4.23.15 #1963 and chore(deps): bump better-sqlite3 from 12.11.1 to 13.0.3 #1058: package.json (1 line) plus pnpm-lock.yaml.
  • chore(deps-dev): bump fumadocs-core from 16.9.3 to 16.15.14 #1962: those two, plus apps/docs/{package.json,pnpm-lock.yaml}, apps/docs/app/[lang]/layout.tsx (12 labels re-keyed one for one to fumadocs-ui 16.15's i18n keys, with satisfies Partial<Translations> so that a stale key is a type error), and one comment line in test/helpers/heading-label.ts (the claim amendment above).
• CI on all three heads: every check is green or skipped; mergeable_state clean.

The dev's measurements, accepted:

• Lockfiles: each regenerated with pnpm install from main's lockfile. Each moves only the bumped package's own tree.
• chore(deps): bump better-sqlite3 from 12.11.1 to 13.0.3 #1058: the installed @objectstack/* 17.6.0 packages ask for better-sqlite3 ^13.0.3. A boot on SqlDriver(better-sqlite3) returned REST create 201 and read 200, and the process mapped prebuilds/linux-x64.node. So the module loads, not only installs.
• chore(deps-dev): bump fumadocs-core from 16.9.3 to 16.15.14 #1962:
  • Root cause: the repo's own pin guard (test/docs-anchor-links.test.ts: root fumadocs-core must equal apps/docs's). The remark-heading slug rule is byte-identical between the two versions.
  • Fix: the lockstep fumadocs-ui move, since its peer pins fumadocs-core exactly.
  • Proof: the docs build makes 359 pages; zh-Hans and zh-Hant chrome is measured in prerendered HTML; zh search answers 200.
  • Maintainer ruling on the upstream fumadocs-ui UI refinements: 「照常合入(推荐)」.

Landing:

• Authority: 「hotcrm一大堆pr你全部接手处理」, and 「授权你执行pr合并」 at about 04:52Z.
• Order: chore(deps-dev): bump tsx from 4.23.12 to 4.23.15 #1963 is enqueued first. Each remaining PR then needs main merged again and the lockfile regenerated, because all three rewrite pnpm-lock.yaml. The seat re-dispatches that refresh, and then enqueues chore(deps): bump better-sqlite3 from 12.11.1 to 13.0.3 #1058, then chore(deps-dev): bump fumadocs-core from 16.9.3 to 16.15.14 #1962.

---

Generated by Claude Code

[objectstack-fleet]

Done: all three dependabot PRs merged through the queue. repo:hotcrm seat, session_01ER8ntXZhYebyQ66aXWdjfT, 2026-10-03T05:17Z.

PR	bump	merged as
#1963	tsx 4.23.15	2e3a8b4d
#1058	better-sqlite3 13.0.3	a1c2c9d7 (refreshed to 2d55f298 after #1963)
#1962	fumadocs-core / fumadocs-ui 16.15.14 + the docs-app i18n re-key	d4a3cafd (refreshed to 065a7b3f after #1058)

• Each head was green on CI before it was enqueued. Each refresh regenerated the lockfile with pnpm install; none was edited by hand. No PR was force-pushed.
• The landing authority is recorded in the ACCEPT 5965721840.
• pm:dispatched and the assignee are cleared in this act.

---

Generated by Claude Code