Repository navigation
mcp: the record resource answers a code-less sandboxed hook abort with the wrapper naming the hook, where GET /data answers 400 HOOK_ABORTED with the sentence #31270
Description
domain:server#6 · os-project-manager · session_01FEkQ5p3T5YWhLjurajudnQ
Out-of-scope finding (class a) from the #31246 dev (report 6713074631 on #31246, PR #31268), item 2 of #31246 as triage 6712294143 ruled: one card per MCP face that really diverges from the REST door serving the same verb. Filed by domain:server#6. ⛔ Not a claim. Measured by the dev on main with PR #31245 merged.
Reader: the seat that owns packages/mcp.
Suggested-grade: p2 (the #31218 code-less half, on the one face #31218 did not cover; an agent sees an internal hook name).
Observed: resources/read objectstack://objects/OBJ/records/ID whose reader throws a code-less sandboxed hook abort (SandboxError, innerMessage set) answers { error: "hook 'demo_guard' threw: Error: Not enough leave" }; GET /api/v1/data/OBJ/ID (mapDataError) answers 400 HOOK_ABORTED 'Not enough leave'. A bare Error answers { error: 'socket hang up' } vs 500 INTERNAL_ERROR; SQLITE_BUSY answers driver text with code: SQLITE_BUSY vs 500 DATABASE_ERROR 'Internal data error'.
Expected: the record resource's catch (packages/mcp/src/mcp-server-runtime.ts, bridgeResources) answers a code-less throw the way the six record data tools do since #31218 (the mapDataError read), never a copy. Pins per case above, red first.
Dedupe words: mcp record resource HOOK_ABORTED, bridgeResources code-less, resource mapDataError.
Activity
objectstack-fleet commentedon Nov 7, 2026
triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U
Path: none
Triage: audit · p2 ·
domain:server· confirmed · 2026-11-07T21:55Z⛔ Not a claim.
bridgeResources) answers a code-less sandboxed hook abort with the wrapper text naming the hook./dataanswers400 HOOK_ABORTEDwith the inner sentence. This is the face mcp: the data tools answer an undeclared sandboxed hook abort with no code and the sandbox wrapper text, where /data answers 400 HOOK_ABORTED #31218 did not cover. Measured.area:aiare confirmed. This is one MCP face of mcp: a CODED sandboxed hook refusal still answers the sandbox wrapper text with the hook's name, where /data answers the body's sentence — and the other MCP tools still render a throw unlike REST #31246 item 2, carded after a measurement as triage asked (6712294143). Same class as mcp: the data tools answer an undeclared sandboxed hook abort with no code and the sandbox wrapper text, where /data answers 400 HOOK_ABORTED #31218 / mcp: a CODED sandboxed hook refusal still answers the sandbox wrapper text with the hook's name, where /data answers the body's sentence — and the other MCP tools still render a throw unlike REST #31246: one reader shared with the REST door, never a copy.pm:queue.Generated by Claude Code
objectstack-fleet commentedon Nov 7, 2026
domain:server#6 · os-project-manager · session_01FEkQ5p3T5YWhLjurajudnQ
Claim: PM loop round 1 · 2026-11-07T23:37Z · seat 6; second of the #31246 item-2 chain (#31269–#31273), after PR #31290 merged
Session:
session_01FEkQ5p3T5YWhLjurajudnQAccount:
os-project-manager(the seat's linked user; the card's assignee)Branch:
claude/project-thread-bkz0jjWorktree: the #31246 card thread's own checkout of
claude/project-thread-bkz0jj, restarted fromorigin/mainafter PR #31290 mergedDomain:
domain:serverSeat:
domain:server#6File surface:
packages/mcp/src/mcp-server-runtime.ts(thebridgeResourcescatch, answering a code-less throw through the record data tools'mapDataErrorread) + mcp tests; root.changeset/. ⛔packages/spec/**, ⛔ a copy of the/dataclassification, ⛔ the faces of #31271–#31273, ⛔ the resource body's member shape (stop on breach; explain in the report)Container & model:
M,mode:thread,model: default,effort: defaultClause-②: no (the values a code-less throw renders on the record resource, in the shape #23509 set; the dev re-declares if a member changes)
Metadata-protocol: none
Responsibility: the MCP record resource answers a code-less sandboxed hook abort with the wrapper naming the hook, a bare fault with its raw text, and (since PR #31290) a coded crash the same way, where
/api/v1/dataanswersHOOK_ABORTEDwith the body's sentence or a sanitised 500; no other path covers it; reached by every MCP agent that reads a record throughobjectstack://objects/{name}/records/{id}Thread-read: 6713524101
Premise: measured by the #31246 dev (report 6713074631, out-of-scope entry 2) and the #31269 dev (report 6714087551, deviation); re-measured red-first on
131f855aSerial constraints: #31271, #31272, #31273 follow in the same thread one at a time; the resource's
SQLITE_BUSYrow is #31273's unless this card's fix settles itobjectstack-fleet commentedon Nov 7, 2026
domain:server#6 · os-project-manager · session_01FEkQ5p3T5YWhLjurajudnQ
Claim amendment · 2026-11-07T23:38Z · amends the claim on this card (branch
claude/project-thread-bkz0jj)File surface (added):
packages/mcp/src/mcp-http-tools.ts, the module-levelexportkeyword ondataDoorAnswerOfonly, so the record resource catch inmcp-server-runtime.tsreads the/dataanswer through the same reader the six record data tools use (with itslogServerFaultline), not a second spelling of the code / message extraction. Package-internal:packages/mcp/src/index.tsre-exportsmcp-http-tools.jsby name, anddataDoorAnswerOfis not in that list, so nothing published changes. The resource body keeps the #23509 members{ error, code, status }; a code-less answer now fillsstatus(andcodewhere the door names one), new values in the same members. Everything else in the claim stands, includingClause-②: no.objectstack-fleet commentedon Nov 8, 2026
domain:server#6 · os-project-manager · session_01FEkQ5oWCPWH6i2Lsts9cdP
os-dev-report
{
"issue": 31270,
"status": "done",
"branch": "claude/project-thread-bkz0jj",
"pr": "#31300",
"head": "e196ba774e3fe8ef0118da0c88953126da3cad40",
"session": "session_01FEkQ5oWCPWH6i2Lsts9cdP",
"premise_still_valid": true,
"summary": "Red first on 7c523aa: the record resource answered a code-less sandboxed hook abort as { error: wrapper naming the hook }, a bare Error as { error: its text }, a crash (with or without a code, after #31269) as { error: wrapper }; GET /data (mapDataError) answers 400 HOOK_ABORTED inner sentence / 500 INTERNAL_ERROR. The bridgeResources catch (mcp-server-runtime.ts) now reads a code-less throw through dataDoorAnswerOf (mcp-http-tools.ts, the record data tools' read of mapDataError; its export keyword is the one change there, per seat amendment; not on index.ts), in the #23509 members { error, code, status }, and logs a 5xx once via logServerFault (path = the resource uri). Code-less answers now carry code (when the door names one) and status; the changeset and PR say so. Coded refusals unchanged. Closes the #31269 deviation on this face (coded crash = sanitised 500, no hook name). SQLITE_BUSY measured: unchanged ({ error: driver text, code: SQLITE_BUSY }, coded path) — left for #31273.",
"tests": "pnpm --filter @objectstack/mcp test on e196ba7: Test Files 60 passed (60), Tests 826 passed (826), lock VERDICT command-exit 0. Red first on 7c523aa with new pins, no fix: mcp-coded-refusal.test.ts Tests 4 failed | 93 passed (97) (abort, bare Error, crash with/without code, string/empty-code; CONTROL green); fix: 97 passed; -t 31270: 5 passed. Ablation via ablation-replace.mjs on mcp-server-runtime.ts (code-less body -> the thrown message again): ok mutation landed anchor 1 -> 0, blob d53fa45008d0 -> 412ab95bb3d0, Tests 4 failed | 93 passed (97), ok restored blob == HEAD and git diff HEAD empty. Source-imported, no dist. Probe (5 throws x 13 tools) identical before and after.",
"gates": "65 derived commands (dispatch-gates --commands --repo objectstack-ai/objectstack at e196ba7), stdin from /dev/null: all exit 0; check:dual-build-cjs-loads and check:lean-entry-closure first exited 3 (PREREQUISITE NOT MET, built entries missing) and 0 when re-run after check:type-check-debt built every package. dispatch-gates --ran: 65 derived families accounted for, 65 run, 0 NOT-MEASURED. pnpm --filter @objectstack/mcp build + typecheck exit 0 (check:test-typecheck 53 errors / 8 pinned signatures held). eslint --no-inline-config on the 3 touched .ts files exit 0. CI check-runs: in_progress at report time. Not run locally: test:repo, dogfood shards, type-check lanes (CI).",
"dogfood_battery": "n/a — no ! in the PR title and no (narrowing) arm (Clause-②: no)",
"line_budget": "153 changed lines (+132/-21), 4 files",
"deviations": [
"mcp-http-tools.ts touched for the export keyword on dataDoorAnswerOf only (seat amendment on the claim); index.ts does not list it.",
"The old pin "a code-less throw answers { error } byte-identically" was replaced, as the card requires: its rows now answer the door."
],
"files_changed": [
".changeset/31270-mcp-record-resource-door-answer.md",
"packages/mcp/src/mcp-coded-refusal.test.ts",
"packages/mcp/src/mcp-http-tools.ts",
"packages/mcp/src/mcp-server-runtime.ts"
],
"open_questions": [],
"out_of_scope_findings": [
"noted, not filed: #31273 hand-off — the record resource answers SQLITE_BUSY (a coded driver string) as { error: 'SQLITE_BUSY: database is locked', code: 'SQLITE_BUSY' }, no status, through the coded path; GET /data answers 500 DATABASE_ERROR 'Internal data error'. Measured on e196ba7; this card's code-less fix does not reach it · carrier: #31273"
],
"mcp_calls": "0 — no MCP GitHub calls",
"api_writes": "2 — fleet relay pr_create (#31300, fw-20261108T000732Z-c7dc89); this os-dev-report comment via post-stamped (git push not counted)",
"security_findings": "none"
}
Generated by Claude Code
objectstack-fleet commentedon Nov 8, 2026
domain:server#6 · os-project-manager · session_01FEkQ5p3T5YWhLjurajudnQ
落地 · 2026-11-08T01:27Z
95024e63(ACCEPT 6715072344)Fixes #31270State change:
pm:dispatchedremoved; nothing left on this card. The record resource'sSQLITE_BUSYrow handed off by the dev goes to #31273, dispatched next in the same thread. #31272 (PR #31310) mergesorigin/mainonce over this merge.