- Drawer tabs 当前/历史: searchable run list (status/source/time filters,
cursor auto-load), opens runs inside the drawer, keeps scroll and filters,
remembers the last tab; 在页面打开 is secondary
- Ended runs: append to resume, or 用同目标新建 with 先看方案再执行 toggle
- Session page: 88ch conversation column, turns grouped with one timestamp,
tool cards collapse to one line, 3+ activities fold; steps/subagents/budget
moved to a right rail (collapsible overview bar on narrow screens and in
the drawer), long model reports fold behind 展开全文
- Smoke covers the drawer history flow
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Auto-fix of 602 string literals across 74 files (half-width punctuation next to
Chinese, parenthesised Chinese, quotes); biome reflow; e2/e6 assertions follow
the new full-width summaries. check:copy now runs with --strict in ci.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
- session page shows the agent.plan TodoList (drawer: one-line progress); options and approval
cards sit at their step; composer has a "先看方案再执行" toggle instead of the mode switch
- settings expose option timeout, timeout action and approval validity; quick goals follow glossary
Co-Authored-By: Claude Opus 5.5 <[email protected]>
pi.provider / baseUrl / model / thinkingLevel / apiKeySecretRef are now
DEPRECATED_SETTINGS_FIELDS: kept in the schema for stored JSON, dropped
from PATCH /settings and from the settings sections, never read by the
server. runAgent fails with a clear message when no upstream is set.
The agent demo gate checks the default upstream provider; e2e seeds a
fake upstream as default. OpenAPI summaries say Agent upstream.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
- agent_run gains depth / spawn_reason (0029); agent_run_event types
subagent_spawned / subagent_event / subagent_completed
- agent.spawn (propose, medium, replay never): allowlist ⊆ parent, mode not
above parent, budget carved from parent's remainder, idempotent per run+input;
settings.pi.subagents {enabled, maxConcurrent, maxDepth} (subagentsEnabled kept as alias)
- child runs claim a concurrency slot under an advisory lock (waitForFact while
queued); agent.await polls business tables without calling the model
- parent abort cascades to all descendants; parent runtime budget counts
descendant usage; child events mirrored into the parent stream by parent seq
- authorize caps every child call by all ancestors' allowlists/kinds
- session page shows parent link, inherited info and the subagent tree
- e2e E11; runbook section on subagents
Co-Authored-By: Claude Opus 5.5 <[email protected]>
- Port agentui (MIT, ashish200729/agentui@b2b752b) Message/MessageBubble/MessageScroller,
StreamingResponse, ToolResult/ToolApproval, TodoList, ApprovalCard, CodeBlock/FileDiff,
Citations, PromptInput into components/agentui; drop motion/shiki/Next, switch colors,
type scale, radii, durations and easing to semantic tokens, honour reduced motion
- Conversation: streaming cursor, collapsible thinking, tool cards with kind/scope/risk/
idempotency badges and audit idempotency key, prominent server-rejection state for both
whitelist rejections and unauthorized errors, compaction markers, budget meters with
over-limit state, approval cards (admin session only), two-column final report
(model said vs system verified with task/record links), follow-scroll with jump-to-bottom
- Composer: Enter send / Shift+Enter newline, stop = abort (confirmed), append,
execute/plan_only switch, "execute this plan" from finished plan_only runs
- Session list sidebar with status filter and usage; reconnect/error banners with retry
- Drawer: route-derived context (entity/site/task), minimize to a live mini dock
- Fix stream refetch race when the run switches from loading to finished
Co-Authored-By: Claude Opus 5.5 <[email protected]>
- POST /approvals and PATCH /approvals/{id} require a console admin session
(EndpointDef auth:"session"); API keys incl. admin scope get 403, so an
organize.execute key can no longer approve its own plan (S1)
- tests for single-use approval checks: expired, revision/payload hash
mismatch, consumed, revoked, pending/rejected, plan mismatch; re-approve
after plan change succeeds once and cannot be replayed (S4)
- one redact key table in @umt/contracts shared by server logger and agent
redactForModel: rsskey, torrent_pass, authkey, downhash, c_secure_pass,
passkey variants, x-api-key, cookie headers, nested objects/arrays (S2)
- TMDB network errors: message and cause redacted via redactUrl (S3)
- web Agent ?mock=1 only in DEV builds or when pi.provider is fake (S7)
- EndpointDef auth annotations for /auth/*, asserted against
PUBLIC_AUTH_PATHS (S10); doc drift fixes (S11)
Co-Authored-By: Claude Opus 5.5 <[email protected]>
- Agent page uses contracts (AgentRun/AgentRunDetail/AgentGoalReport) instead of a local
mirror; ?mock=1 stays as an explicit demo switch; run snapshot refreshes on agent.run_status;
stream auto-follow only while the user stays near the bottom (abort stays reachable)
- settings: Pi provider select incl. fake, new Agent runtime card (budgets, fakeScript)
- download trace uses /tasks?subjectId= and /organization-records?downloadTaskId= instead of
time-window engine run matching and path matching; new business-task tab with cancel/retry
- workbench download dialog posts /acquisition-plans and shows per-slot rejection reasons
- preference tag revision history; subscription search by title/kind; identity/alias/subscription
lists show entity titles; slot cards show current asset summary; policy preview sample titles
- DecisionSummary / IdentityRevisionList / ReleaseIdentityResult / TaskAccepted / PolicyPreviewSummary /
OrganizePlanArtifactContent taken from contracts
- docs: SSE for console (cookie) vs GET /events?after=&limit= JSON replay for API keys
Co-Authored-By: Claude Opus 5.5 <[email protected]>
- sites + /sites/:siteId: organized view with snapshot-first release list, inventory overlay,
candidate reasons, volatile-field recheck, stale marking that keeps existing data, live search,
unified download entry; origin view placeholder (M2)
- sources: feed CRUD/enable/fetch/reset, cursor/health/gap hints; per-site snapshot history
- tasks: download tasks with pause/resume (threshold pause cannot be bypassed), plans,
cache volume thresholds and cleanup, business timeline (observation → plan → download →
organization → agent/engine)
- organize: record list/detail with file-level results, manual organize with dryRun preview
- agent: session list/page and layout drawer with context prefix; streaming events,
tool cards, rejection alerts, compaction, budget meters, model text vs server report;
mock switch (?mock=1) until W2-D endpoints land
Co-Authored-By: Claude Opus 5.5 <[email protected]>
