Repository navigation
Description
domain:server#1 · sunhaolin · session_017xRBcTJTE7kSbcMoJQ4Txf
Filing-gate category: ① a named-location documentation defect, owed as the Skills-docs: follow-up of PR #31083 (#31021).
Why (business)
An author or an AI agent learns how seeds behave from these two pages. After #31021 lands, a seed row a person deleted is not inserted again on the next replay. The docs do not say so, so a customer who wants the sample data back has no documented way to get it. Separately, the published objectstack-data skill teaches a default seed mode and a replace behaviour that have not been true since #30081, so an AI that follows it writes seeds that do something other than what it expects.
Items (two lanes; triage routes or splits)
content/docs/data-modeling/seed-data.mdx: teach the landed-keys rule from PR fix(metadata-protocol,runtime): a seed row a person deleted is no longer inserted again by the next seed replay (per-organization record of landed seed keys) #31083. Each organization records which seed rows landed; a replay (install, upgrade,serveboot) does not insert again a recorded row that is now gone; a deliberate re-seed clears the record (seed-replayer(orgId, { reseed: true }), per the PR). Source: the PR fix(metadata-protocol,runtime): a seed row a person deleted is no longer inserted again by the next seed replay (per-organization record of landed seed keys) #31083 changeset and body. The page already describesignoreas the default correctly (:74,:80).skills/objectstack-data/references/seeds.md, the "Import modes" table: it saysupsert(default) andreplace= "Delete everything, then insert. Data loss." Onmain,ignoreis the default ([Decision] A restart puts seeded records back: a customer's edit to a seeded row is overwritten by the next boot #30081, andseed-data.mdx:74), andreplacedeletes nothing (seed-data.mdx:78: "the seed loader does not delete anything itself"). Align the table with the docs page and the loader. A skills line budget applies.
Reader: item 1 is docs (content/docs/**); item 2 is the published skills (skills/**). Item 1 waits for PR #31083 to merge.
Blocked-by: #31021
Dedupe (2026-11-07): seed-data.mdx deleted seed row → 10, seeds.md mode table → 7 (top hits: #31021 itself and closed skills cards #26297, #30857, #26475, #22149, #26184, #19715; none covers either item), seed mode default ignore docs → 374, landed seed keys docs → 687 (too broad to read; the first two queries are the discriminating ones).
Activity
objectstack-fleet commentedon Nov 7, 2026
domain:server#1 · sunhaolin · session_017xRBcTJTE7kSbcMoJQ4Txf
Item 3 (spec lane), from the contract review of PR #31083 (6702871453, should-fix S1).
packages/spec/src/data/seed.zod.tspublishes the meaning of seedmode(its docstring and.describe()text: "a missing row is inserted and an existing row is left as it is") as the unconditional create-only rule. After #31083, AppPlugin's loads do not insert again a row whose key landed in the organization and is now gone, and the deliberate re-seed is theseed-replayerservice with{ reseed: true }. Amend that text beside item 1. It is apackages/specfile, so the spec lane edits it; the skills or docs lane does not.objectstack-fleet commentedon Nov 7, 2026
triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U
Path: none
Triage: audit · p2 ·
domain:skills·pm:blocked· 2026-11-07T05:53Z⛔ Not a claim.
seed-data.mdxteaches the landed-keys rule, under which a deleted seed row stays deleted until a deliberate re-seed. Item 2: the publishedobjectstack-dataskill's "Import modes" table still callsupsertthe default andreplace"Delete everything". Onmain,ignoreis the default andreplacedeletes nothing ([Decision] A restart puts seeded records back: a customer's edit to a seeded row is overwritten by the next boot #30081).domain:skills, with item 1's docs page folded in and a cross-seat note forcontent/docs/**. One PR keeps the two pages in step.pm:blockedon the card's ownBlocked-by: #31021. PR fix(metadata-protocol,runtime): a seed row a person deleted is no longer inserted again by the next seed replay (per-organization record of landed seed keys) #31083 is in the merge queue now (group running at 05:42Z), so the wait should be short. If it stalls, the seat may land item 2 first and leave item 1 on the card.Generated by Claude Code
objectstack-fleet commentedon Nov 7, 2026
domain:skills#3 · huangyiirene · session_01J8VVAj7CvjkGXvgXzd6RZ7
Unlock:
Blocked-by: #31021is met. #31021 closedcompletedat 2026-11-07T06:02Z, and its PR #31083 merged at 2026-11-07T06:02Z.pm:blocked→pm:queue. Item 3 (thepackages/specseedmodetext, 6702892720) is spec lane; the seat routes it on a separate card and does not ride it here.objectstack-fleet commentedon Nov 7, 2026
domain:skills#3 · huangyiirene · session_01J8VVAj7CvjkGXvgXzd6RZ7
Claim: PM loop round 14 · 2026-11-07T06:53Z
Session:
session_01J8VVAj7CvjkGXvgXzd6RZ7Account:
huangyiireneBranch:
claude/project-thread-ba8x76(the card thread's designated branch; dev sessionsession_01J8VVAkA4qny8vRgP2NzRgs)Worktree:
objectstack-issue-31090Domain:
domain:skillsSeat:
domain:skills#3File surface:
skills/objectstack-data/references/seeds.md(whole-file sweep of the stale mode claims); hits inskills/objectstack-data/SKILL.md,rules/*.md,evals/*.jsonthat teach the old mode rule;content/docs/data-modeling/seed-data.mdx(the landed-keys rule, cross-seat note forcontent/docs/**);.changeset/*.md;scripts/check-skills-token-ratchet.mjsonly by standing ruling 6513306068 for a measured remainder (stop on breach)Container & model:
M,mode:thread, project-default model,effort: high.dispatch-gates --tieris run by the thread and reported; any mandate above this thread's is waived by the maintainer's standing order in this project (2026-11-03) and said so in the PR body; this seat runs the contract review.Clause-②: no
Metadata-protocol: none
Responsibility:
n/a — not a defect card(docs follow-up of PR #31083 and #30081)Thread-read: 6703573368
Premise: holds on
mainb6315733:seeds.mdstill saysupsert(default) andreplace= "Delete everything";seed-data.mdxdoes not teach the landed-keys rule.Serial constraints: none. Item 3 (
packages/spec/src/data/seed.zod.ts) is spec lane and does not ride this card; the seat routes it.Changeset
'@objectstack/skills': patch.Fixes #31090. Tier read bycheck-governed-mergesat the PR.objectstack-fleet commentedon Nov 7, 2026
domain:skills#3 · huangyiirene · session_01J8VVAj7CvjkGXvgXzd6RZ7
ACCEPT → PR #31097 comment 6704175289, head
b7399145. Tier H:pm:awaiting-merge, approval requested fromos-zhuangandhotlong. Item 3 and theos diffstring → #31099.