|
|
|
@@ -6347,3 +6347,9 @@ R5 quote fallback room (#797 D3, 2026-11-06, MVP3 lead): r4 review P2 found that
|
|
|
|
|
2026-11-09 UTC, PRD 1.20 and defect l (#1873, #1886; PRs #1887 03fe321b1 and #1888 7a97cfc4f). PRD 1.20: Jason approved draft 2 at 22:57Z ("The file looks fine other than that"; the Montage module name stays deferred). The PRD now has 118 requirements, with REQ-WS-1..8, REQ-AGENT-1..3, REQ-SYS-1..2, REQ-ACCT-1..2 and REQ-SETUP-1..2 added and REQ-ASSIST-1, REQ-USER-1, REQ-TASK-1 and REQ-WEB-3 amended. Builder Codex Terra xhigh declined to commit because stack-copy failed 20/1 (node-gyp missing from the test Node's npm); that failure is pre-existing on 483eccb3d, so the lead committed. Sol review APPROVE 0 (word-for-word diff against draft 2, catalog order, both requirements tests). Defect l: before, an agent that declared a store credential made the next bus start refuse, because Credentials refuses a missing store root and only issue() created it. After: start creates the root 0700 when any reference names the store; a root that exists with loose modes or as a symlink still refuses. Two new credential-verbs tests fail on the parent. Bus suite 171 files one at a time, 1713/0 (Node 26.12.1, env -i). Sol review APPROVE 0. uiorg redeployed at 7a97cfc4f, 23:15Z. Follow-up: the stack-copy node-gyp test environment. Reports: /mnt/storage/scratch/mosaic-lead/reports/{prd120-land,prd120-review,defectl-review}.md.
|
|
|
|
|
|
|
|
|
|
2026-11-10 UTC, unified workspace U1 and SYS1 (#1873; PRs #1889 d35447005 and #1890 2c6556ab1). U1, before: each org page read only its own bus and nothing read every org at once. After: /api/workspace returns projects, sessions, inbox, tasks and metrics from every org the owner holds, each row tagged with its org and each org with a state (ok, starting, down, timeout, invalid); `mosaic console restart` and `mosaic org reload <org>` send the F3 routes. Sol review CHANGES 0/2 on 2109458b3: malformed data from one org turned the whole read into a 503, and the per-org deadline left sockets and transport children running. Terra fix 0672d5fe8 validates each org's result inside its boundary and carries an absolute deadline and AbortSignal through the status check and transport; the new workspace-failures test is 0/10 on the parent and 10/0 on the fix. The lead re-pinned the webui bus.mjs hash in bus channel-proof (the deliberate signal/deadline change; nonces, argv and limits unchanged) and amended as d35447005. SYS1, before: no system-type kind. After: a system-type plug-in kind with five modules (business-management, coding, creative-writing, marketing, montage), an org enable list under plugins.system-type, owner-only project selection with an append-only project_system_types table (migration 3t-3u) and CLI verbs; the modules grant no tools yet (REQ-SYS-2 deferred). Sol review CHANGES 1/2 on a9629f332: the restore guard did not count project_system_types, the CLI channel did not admit the two project verbs, and the #811 migration-reader pin still expected 18 steps. Lead fix 2c6556ab1 (rebased on d35447005) adds the table to COUNTED and LISTED with refusal and in-run recount tests, admits both verbs with a run()-level channel test (red on the unfixed channel), and moves the pin to 19 steps ending 3t-3u; fix files 323/0 and rebase overlap files 63/0 (Node 26.12.1, one file at a time, env -i). Full gate on a9629f332 4040/5: the two stack-release pins above, and tester-reads 3/3, which also fails on 1616d1dd6 (project.mjs:95 refuses an org --tester init; follow-up). Correction to the uiorg deploy at 00:50Z: the deploy script switched the tree without the offline schema migration, so all five buses refused to start (normal startup refuses an older schema by design). The lead ran migrate-schema.mjs for the five org stores (3t to 3u, all ready), restarted the buses (201 each) and added the migration to the deploy script with stop-on-refusal. test3 refused writer-locked and stays down, as before. Reports: /mnt/storage/scratch/mosaic-lead/reports/{u1-review,u1-fix,org-sys1-build,sys1-review}.md.
|
|
|
|
|
|
|
|
|
|
2026-11-13 UTC, starter packs replace system types (#1873; PRs #1949 c0a60ed3a, #1950 1e65ff3f8, #1951 755ba3fa8, #1952 1ced0e392, #1953 dfef78834, #1955 9a649abf8, #1956 991eb8c33, #1957 34dd33101, #1959 0251acdad and #1960 e6ab8bc1f; final e6ab8bc1f). Backfilled 2026-11-16 by lane docs-1116 from uidev/LEAD-1110.md and git; the Lead wrote no entry at the time. Before: SYS1 (PR #1890, 2c6556ab1) made system types a plug-in kind with five modules, an org enable list under plugins.system-type and an append-only project_system_types table (migration 3t-3u); the modules granted no tools. After: the design docs/plans/2026-11-12_starter-packs-design.md landed as PR #1949 (doc only) after three fix rounds; Jason's FULL SEND at 21:06Z on 11-12 answered Q1 to Q3: system-type removed outright with no legacy, SP10 dropped, SP9 deletes the kind and drops project_system_types. Slices in landing order: SP3 the org keys packs, board and workflows and the applyPack writer op; SP1 the starter-pack plug-in kind and its checker; SP7 workflow guidance in the harness prompt; SP2 the five shipped packs; SP4 pack, board and workflow verbs, routes and CLI words; SP6 board labels, pack picker and workflow editor on the WebUI organization page; SP5 the chief applies a pack's non-definition half and raises hires from its proposals; SP8 the owner saves an organization's shape as a template; SP9 system types removed outright and bus schema 3v-3w dropping project_system_types. Test counts, Lead check then lane gate: SP3 98 files 1294/0, 1828/1; SP1 51 files 708/0, 2854/1; SP7 1281/0, 2537/1; SP2 150/0, 1556/1 at 18add86cb before its rebase onto SP7; SP4 4848/2, 6049/1; SP6 371 files 4233/1, 8262/23; SP5 229 files 2721/0, 8061/2; SP8 415 files 4650/1, 8125/2; SP9 497 files 5542/1, 8655/28. Each single Lead check red is CHAT-01 (host Python lacks jsonschema), red on base too; SP4's second red was the fixture federation-host.mjs wrongly in the run list, and Lead check lists are filtered to *.test.mjs since. The lane gate reds the log names are base or environment reds: CHAT-01, 19 agents/*/work probes, budget-supervisor 3v, ui-browser SSO, six v1 .woodpecker files and one claude-session timing test. Every slice was deployed to uiorg 7350 and verified; at e6ab8bc1f (03:46Z on 11-13) the offline 3v-3w migration ran on all five live stores, 5 stop 200, 5 bus schema 3w ready, 5 setup 201, the system-types route and its js returned 404, and no org file named system-type, so no wipe was needed. Corrections: the Lead's 21:38Z carry that SP2 pack steps must drop id was reversed at 21:46Z: pack steps keep id (SP1 checker) and SP4 drops id when it copies a step into the org file. The Lead's SP7 brief misstated the workflow shape; the lane built to the SP3 loader, which was correct. SP5's last commit before landing changed the broker canonical string builder after its re-check with no suite test; the Lead compared it with organization canonicalJson (7/7 same) and accepted it, and the integer-key pin test is deferred. SP9's first gate, before its rebase, ran packages/cli/live/ui-browser.test.mjs, which binds its own fixture on 17443, a lane rule breach; nothing was listening before or after, and the file was left out of later gates. The 'bus recover' deploy step refuses rc=3 in the agent shell on every deploy since D1; it is harmless while bus stop succeeds. Reports: /mnt/storage/scratch/mosaic-lead/reports/{starter-pack-design-build,starter-pack-design-recheck2,sp1-build-1112-starter-packs,sp2-build-1112-starter-packs,sp3-build-1112-starter-packs,sp4-build,sp5-build,sp6-build,sp7-build,sp8-build,sp9-build,sp9-rebase-recheck}.md.
|
|
|
|
|
|
|
|
|
|
2026-11-15 UTC, applying a starter pack is the hire decision, with the route fallback (#1873; PR #1963, b913f0f4c; commits 359b3299b and 3e53af394 lane pd, bfd2cc5e7 and b913f0f4c lane pr2). Backfilled 2026-11-16 by lane docs-1116 from uidev/LEAD-1110.md and git. Before: section 3 of the starter packs design said apply never hires; a person's apply left the pack's hires as proposals for the chief to raise or for a person with org.hire to start. On 7350 at 65f7288ce Jason applied packs and saw no agents. His ruling at 20:48Z on 11-14, verbatim: "Damnit. If I choose a starter pack and apply it, that is a decision!" After, lane pd: a person's apply (pack.apply.console from the WebUI, pack.apply from the CLI) raises one hire decision per proposal after the pack's commit and resolves it, with the evidence an inbox approval leaves. The owner's apply approves flagged and credential lines; a member with org.hire hires only proposals with neither, and the rest wait for the owner as awaiting-owner. An agent apply hires nothing. Hires are written with no credential material, so services show unbound until someone binds them. The design's section 3 carries a 2026-11-14 change note. Route fallback: at 22:03Z the Lead found every 7350 chief had model null, so pack hires would refuse pack-hire-route-missing on Jason's orgs. Ruling at 22:13Z: the hire route is the chief's, else the org assistant's harness and model, else refuse; lane pr2 stacked on pd. Tests: pd gate 3957/0 at its first commit, review changes 6 (2 P2) fixed or deferred, re-check APPROVE; Lead check of pd with pr2 333 files 3813/1, the red a control-board scan.test mtime-tie flake in a file neither lane touched, 70/0 three times alone. Lead ruling on the re-check's P2 1: a member pack hire surface is a follow-up row, because a new transport for a member hire cap is a security boundary and the owner path covers apply; Jason said "proceed". Deployed b913f0f4c at 01:03Z on 11-15: stop 5, schema 3w 5, setup 5. The Lead ran pack hire on faceless (montage ideas-1 and research-1, coding pm-1, coder-1 and reviewer-1, all on fixture-deterministic) and acme (business-management pm-1 and research-1 on opus), each with routeFrom assistant. Related fix the same night: Jason reported "adapter exited 127" in Messages. Lane r127 found the Lead's deploy-uiorg.sh (a scratch script, not in the repository) started the Console without MOSAIC_CLAUDE_BIN, so claude-code sessions ran the /usr/bin/claude wrapper whose target under /opt the sandbox does not bind. The Lead patched the script and landed PR #1964 387f73a7a, which makes exit 127 name the missing program; Lead check 16 files 285/0. Both assistant kinds replied on 7350 at 387f73a7a by 03:33Z. Reports: /mnt/storage/scratch/mosaic-lead/reports/{pd-build,pd-review,pd-recheck,pr2-build,pr2-review,r127-build,r127-review,r127-recheck}.md.
|
|
|
|
|
|
|
|
|
|
2026-11-15 UTC, single pane slices sp1 to sp14 (#1873; PRs #1965 to #1978; final c26879637). Backfilled 2026-11-16 by lane docs-1116 from uidev/LEAD-1110.md and git. Before: a Side by side page sat in the sidebar beside the organization pages, and Jason's walkthrough on 65f7288ce (20:48Z on 11-14) found agents from an applied pack not visible. His direction: #/dashboard system-wide, no organization URL base, side-by-side Assistant threads in place of Side by side, and the 3-click rule. The Lead ruled at 03:33Z on 11-15 that #/orgs/<id>/... stays the address of an org page but is a route inside one global shell, with no reload and no mode switch. After, slice by slice (PR, tip, Lead check, lane gate): sp1 shell, #/dashboard system page, org drill-in without reload, a down org says so (#1965, c7aecbffa, 128/0, 1031/0, two reviews, nine findings fixed); sp2 dashboard polish, bare org route, theme, one down banner (#1966, d1b01329b, 132/0); sp3 system Inbox (#1967, 3885f4628, 111/0, 892/0; a theme fix and a render-order fix were read by the Lead, not a reviewer); sp5 system Agents, landed before sp4 (#1968, f97e0d3ae, 89/0 plus unified-workspace 18/0, 745/0); sp4 system Tasks, which the Lead rebased onto f97e0d3ae with three one-line union conflicts (#1969, ed4f45ed2, 131/0, 937/0 before the rebase); sp6 system Assistant with up to three threads from any org side by side, the open set in the hash, Side by side removed from the sidebar and #/side-by-side landing on Assistant (#1970, 789e2bcdd, 95/0, 1472/0, four review rounds); sp7 seven small fixes (#1971, a9912983e, 141/0, 1483/0, three review rounds); sp8 KPI tiles from real reads only and the 3-clicks table (#1972, c66f1ec00, 119/0 plus org-live-edit-browser 4/0 alone, 1490/2 with two load flakes that each passed alone twice, three small fixes not re-reviewed); sp9 read-only workflow list and graph per org (#1973, 7c6571768, 143/0, 1504/0); sp10 workflow edit in the graph (#1974, bd29e756a, 146/0, 1515/0); sp11 eight fixes including agent counts that disagreed and Done this week (#1975, 5b6b39f08, 164/0, 1527/0, three review rounds); sp12 agent drawer from #/agents and org session rows (#1976, 69ce460f9, 196/0, 191 files 1538/0); sp13 organization page fits its width, #/orgs/<id>?query parses, drawer state pending until the instances read answers (#1977, 68df55130, 205/0, 192 files 1547/0); sp14 Queue card off the organization page, running not read where an answer carries no runs, kept Assistant pane messages keyed by an opaque viewer, drawer repaint without new page reads (#1978, commits 2fc97a3ed and c26879637, 218/0, 192 files 1549/0, review changes 1 with the P2 fixed, re-check approve). sp14 adds `viewer` to /api/config (32 hex of sha256 over provider, iss and sub; `local` without sign-in) and adds no route. The Lead ruled at landing of sp13 that the Queue card leaves the organization page; #/queue stays. State at c26879637: system pages #/dashboard, #/inbox, #/agents, #/tasks and #/assistant; org pages at #/orgs/<id>/... inside the same shell. Each slice was deployed to uiorg 7350 with five buses up (deploy log per tip under reports/deploy-<sha>.log); sp1 and sp8 to sp14 record 5 stop, 5 schema 3w and 5 setup. An acme assistant reply was checked live after sp1 and sp6 to sp14; after sp4, #/agents showed the acme assistant running mid-reply; the log records no reply check for sp2, sp3 and sp5. Corrections: sp15 was launched at 19:23Z on a wrong premise. The Lead had told Jason, from reading code only, that the broker's async instances path was the real cause of the sp12 drawer saying not running. The lane stopped at BASE with no diff: the shipped runtime answers instances synchronously (runtime.mjs:1263), and a probe on a real startBroker showed the running and alive arrays present. The sp12 live shot had a 9000 ms screenshot budget that caught the paint before the read; the sp13 shot at 30000 ms shows State running. sp15 closed at 19:26Z with no change and no bus row; sp14's running not read item now only guards a failed or odd read. The Lead's lesson: prove a fault live before naming it a cause. The sp14 live shots read 0 running because they were taken right after the bus restart, before any session started. Follow-ups after sp14: empty space under Active agent sessions at 1440, a drawer pending gap, viewer read once per tab with an unsalted hash, the open pane set kept per browser, and the Dashboard org table scrolling sideways at 390. Reports: /mnt/storage/scratch/mosaic-lead/reports/{sp1-build,sp2-build,sp3-build,sp4tasks-build,sp5agents-build,sp6threads-build,sp7polish-build,sp8kpis-build,sp9workflows-build,sp10workflowedit-build,sp11fixes-build,sp12agentdrawer-build,sp13orgpagefit-build,sp14orgpagetruth-build,sp15businstancesruns-build}.md.
|