Reduce image_storage_bytes from 500MB to 25MB on free and from 1GB to
500MB on pro to align quotas with current storage cost expectations.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Persist host/registrable_domain/site_key on monitoring article URL
aliases, populate them on desktop-publish completion, and use them in
the citation pipeline to count how often SaaS-published content
appears as a citation source. Expose a /citation-summary endpoint with
a 7/30-day window switch and surface the new source-share metrics in
the tracking dashboard.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Replace the legacy target_platform string on schedule tasks with a
workspace-aware auto-publish payload (auto_publish + publish_account_ids
JSONB + cover_asset_url + cover_image_asset_id) and migrate the schema,
sqlc queries, generated models, domain struct, ScheduleTaskService DTOs,
and dispatch worker to round-trip the new fields. PromptRuleGeneration
gains a WithPublishJobService hook so executeGeneration can enqueue an
auto-publish job once the article is ready, and worker-generate wires
the publish-job service in. On the admin-web side, extract
PublishArticleModal's account-card builders into a shared
publish-account-cards module, rebuild GenerateTaskDrawer's schedule
mode around account selection plus a CoverPickerModal, and surface the
new "auto publish" column on ScheduleTaskTab. Shared types and i18n
strings cover the new fields.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
RequireActiveTenantSubscription now optionally takes an AuthRepository
and rejects the request with 40311/user_disabled when the actor's user
row is anything other than active. Wire the repo through the tenant
router, surface the new error code in admin-web's
membershipBlockedErrors set, persist a "disabled" membership status when
markStoredMembershipBlocked sees user_disabled, and add localized copy
on the blocked screen and shell membership banner.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Tighten the analyze prompts (runtime + recommendation platform) so the
model returns 8-10 question-style search queries (哪家好/有哪些/怎么选/
价格/排名/避坑 …) instead of generic tag words, and lift the keyword cap
in normalizeAnalyzeResult from 5 to 10 to keep them. Add tests covering
the new ten-keyword window and the question-like guidance in the seeded
analyze prompt.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
懂车帝 has the same cover-mandatory rule as 百家号; treat them the same
when deciding whether a publish batch must carry a cover.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Logging the full candidate and selected snippet bodies per resolve was
inflating log volume without adding signal we use.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Wire LoginGuard into tenant-api and ops-api login flows: acquire a
permit before checking credentials, record a failure on every invalid
attempt to drive lockouts, and clear counters on success. Tenant-api
now also forwards the request IP into the service so per-IP and
IP+identifier limits actually fire under real traffic.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Introduce a Redis-backed LoginGuard that combines per-IP, per-identifier
and IP+identifier rate limits, exponential lockout on consecutive failures,
and a concurrent-attempt lock so a single subject cannot pile up parallel
brute-force tries. The guard ships with an in-memory fallback when Redis
is unreachable so login protection keeps working under degraded mode.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Add a lazy Redis client constructor and an in-memory fallback for the
refresh-session store, plus tenant-api and ops-api bootstrap that warns
and continues with the lazy client instead of failing. Refresh and
blacklist operations now silently fall back to the in-memory store while
Redis is down so existing sessions stay valid until it recovers.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Verify approve/revoke/manual-bind delete the right tenant-scoped keys via
a recording cache double, so the prompt cache stays consistent with the
KOL subscription state changes shipped in the previous commit.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Approving / manually binding / revoking KOL subscriptions belonged to the
operations console, not the tenant-admin role on tenant-api. Add a fresh
KolSubscriptionService + repository + handlers under the ops module with
its own list/manual-bind/approve/revoke routes, wire a Redis-backed cache
into ops-api so admin actions can invalidate tenant prompt caches, and
delete the tenant-side admin service/handler now that ops-web owns the UI.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Add /kol/manage/packages/:id/self-subscription POST/DELETE on the tenant
side so a KOL can grant themselves access to their own published package
without going through the marketplace approval flow. Reject self-subscribe
through the marketplace with a clear error, expose owned_by_current_tenant
on package responses, and surface a self_subscription block on KolPackage
Response. The admin-web KOL workspace gets 订阅自己 / 取消订阅 entries with
matching messaging in the marketplace detail view.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
a.publish_status only stored a coarse summary so the article list/detail
mis-reported as 失败 when one of several accounts failed. Compute the
effective status via a LATERAL join over latest publish_records per
platform_account_id and bucket into success / publishing / failed /
partial_success. Wire the same expression into list filtering and detail
queries, and let normalizePublishStatus / batch-status helpers retain
partial_success end-to-end.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Open a separate pgx pool for the monitoring database, wire it into the
SiteDomainMappingService, and surface its health in /readyz. Point compose
at the dedicated monitoring-postgres service and align local/dev configs
with the per-instance ports. Also reformat the site_domain_mapping types
with gofmt while the files are touched.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Register site-domain-mappings CRUD endpoints under the authenticated ops
router, add a monitoring_database config block (falling back to the main
ops database when unset) for the service to share the monitoring pool.
Wraps the repository with input normalization (lowercase domain, host-
extraction from URLs, length limits) and audit logging for
create/update/enable/disable/delete actions, ready to be wired into the
ops console handlers.
Add SiteDomainMappingRepository with list/get/create/update/set-active/
delete to back the ops console management UI on top of the now-global
site_domain_mappings table.
Drop tenant scoping on site_domain_mappings now that the citation query
resolves mappings without tenant filtering. Add a unique index on
registrable_domain plus an updated_at column, and switch the dev-seed
upsert to use ON CONFLICT (registrable_domain). Adds a SiteDomainMapping
type to ops/domain for the upcoming console management UI.
Replace the per-row tenant+global LATERAL joins with a single CTE that
materializes filtered citations and resolves domain mappings once per
distinct host/domain/site_key triple. Reduces redundant scans of
site_domain_mappings on large run batches.
Why: avoid showing accounts as not-live during stale-window probes; keep publish gating consistent with runtime view.
- collapse FIRST_PROBE window so initial probe runs immediately
- skip due-probe selection when a probe is already queued/in-flight
- mirror auth-state→health projection on both client report and tenant ingest paths
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The buffer-backed xdb.Searcher is not safe for concurrent use, so
parallel audit appends could race on Search and corrupt internal state.
Serialize v4 and v6 lookups behind separate mutexes so concurrent
callers fall through one at a time.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Add an IPRegionResolver wrapping the ip2region xdb library and attach it
to AuditService so each appended event records both the raw IP and a
resolved region in a new ops.audit_logs.ip_region column. Loopback and
private addresses short-circuit to local labels; missing xdb data or
lookup errors degrade silently so auditing keeps working without it.
The ops console audit view shows the region beneath the IP. Bundle the
v4/v6 xdb data under internal/ops/app/ipregiondata so the resolver works
out of the box, with config paths/env overrides for swapping in updated
data sets.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Free plan now ships with 10 AI points/month and Plus with 1000, replacing
the placeholder zeroes left over when the feature was first wired up.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds /admin-users CRUD on the ops backend, covering listing, plan/role/KOL
toggles, subscription expiry, status flips and password resets, with a
configurable default plan code. The ops console exposes a new top-level
"用户管理" view and reorganises the sidebar so 操作员管理 and 审计日志
move under a "系统设置" group; AccountsView is renamed accordingly.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Phone becomes the required, unique login identifier; email and name turn
optional. Login now accepts either via a single identifier field, refresh
re-checks tenant membership, and the dev seed provides phone numbers.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Server callback now rejects succeeded submissions without a non-empty answer,
or whose answer contains [citation:N] markers without any sources. Desktop
adapters mirror this client-side: kimi returns unknown when sources arrive
without a final answer, and yuanbao returns unknown when the answer has
unresolved citation markers. Yuanbao also unwraps redirect URLs, harvests
citations from more DOM attributes and document URL fields, drops links
pointing back at yuanbao itself, and dedupes overlapping streaming
fragments. Runtime controller forwards only an explicit "succeeded" status
as success; anything else becomes failed.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds an ai_point_usage_logs ledger and a reserve/refund/complete pipeline
that charges AI points for article selection optimize, template analyze
/title/outline, and KOL prompt generate/optimize. Pending reservations
are reconciled when the kol-assist worker and template-assist tasks
finish or fail, so points refund automatically on errors. Workspace now
exposes the AI quota status and a paginated usage ledger.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Introduces ai_points_monthly and ai_point_base_chars on the membership
plan policy, exposes them through TenantPlanAccess and MembershipInfo,
and seeds Pro with a 1500-point monthly allowance across the local,
server, and deploy configs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Stand up an internal ops-api service with operator account, auth, and
audit log subsystems backed by a dedicated migrations_ops migration set.
Wire Makefile targets and the migrate Dockerfile stage so the new schema
ships alongside the existing tenant + monitoring databases.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Daily monitoring tasks were pinned to a single primary client. If that
client went offline the task wedged even when another desktop client of
the same workspace was online and bound to the same account. Drop the
primary-client constraint on the materialized collect rows, and instead
pick a target per-platform from live account/client presence in Redis,
falling back to the DB client_id when the desktop client is still flagged
online. Desktop task lease for kind=monitor now matches by account
ownership in addition to target_client_id, so any client owning the
account can drain the queue.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Add a POST /desktop/accounts/health-reports endpoint that buffers reports in
Redis, dedupes via signature, and publishes change events to a new
desktop.account.health RabbitMQ queue. A sink worker drains the queue, and
the account service overlays runtime health onto database health when serving
desktop accounts. Publish job creation now consults runtime health so a stale
DB row no longer blocks publishing once the desktop client reports it healthy.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Bump the default LLM from doubao-seed-2-0-lite to doubao-seed-2-0-pro
for higher-quality completions on tenant-side LLM calls; keep the lite
id in a trailing comment for quick rollback.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Now that desktop clients can be re-registered with a new id under the
same user, listing/searching publish tasks by client.id missed the user's
older history and the retry path forbade jobs created from a previous
client. Switch the list/count queries to join desktop_publish_jobs and
filter by created_by_user_id, and enforce the same ownership when
retrying via desktop task. Qualify column names with the t. alias so the
join doesn't break the existing ORDER BY clauses.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Generate the desktop client_id deterministically from OS machine identifiers
(IOPlatformUUID / MachineGuid / /etc/machine-id) scoped per
tenant+workspace+user, with a persisted UUID fallback when the OS lookup
fails. The renderer now requests the id over a new IPC bridge instead of
keeping a localStorage UUID, so reinstalls and storage clears no longer
fork into duplicate clients. Server enforces the id as required and
rejects empty/malformed UUIDs. Also harden bootstrap reveal flow and
single-instance exit so a second launch focuses the existing window.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Lets desktop clients fetch tenant images/articles assets via Bearer-auth
without requiring the public asset signature, so stale-signed assets
served from existing articles still load on the desktop side.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- Walk knowledge group children recursively so selecting a parent group
pulls snippets from its descendants too.
- Filter knowledge items to status='completed' across the prompt and
active-snippet queries to avoid leaking in-progress documents.
- Dedupe vector candidates by point id and normalized text, then pick
rerank+vector results with a per-item cap so prompts stay diverse.
- Tag each rendered snippet with a [Kn] reference id for prompt citing.
- Build the rule knowledge query from task params (title/keywords/etc.)
before falling back to rule prompt content so retrieval matches the
user's actual ask.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The group_id condition was attached to filter.Should, which only contributes
to scoring instead of restricting the result set. Combined with the tenant
must condition this allowed points outside the requested groups to be
returned. Append the match into filter.Must so group scoping is enforced.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Wires up Baijiahao (百家号) and Jianshu (简书) as first-class desktop
publish targets, with risk-control prompts surfaced in the runtime
controller and a normalized error message in publish records. Adds
external-link buttons in the publish management table, an asset
format conversion endpoint for cover image compatibility, and
reorders publish-status display priority so failures take precedence
over partial successes.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Trigger an AI optimize panel from a text selection in the KOL prompt
editor, accept a user instruction, and stream a preview before replacing
the selected fragment. Extend the shared editor AI assist panel with
boundary-aware placement so it can be hosted inside scoped surfaces, and
update the backend optimize prompt to honor the user instruction.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Introduce 仿写创作: new list and create views, backend imitation service
and prompt templates, worker task routing for imitation jobs, and one-click
rewrite triggers from question citation sources. Surface source article
URL/title on article list/detail, and restrict failed articles to delete-only.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Bumps refresh token lifetime to 720h with env overrides, and stops
clearing the local session for non-401 refresh failures so users aren't
logged out by transient network blips. Adds a 60s retry timer when a
refresh fails for non-auth reasons.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Pull Kimi's "搜索网页" panel results into the regular citation source
flow, drop the dedicated-citation-panel branch on the server side, and
unwrap Kimi redirect URLs so citation source items resolve to the real
target host.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Skip article alias resolution and content-citation aggregates when the
platform is qwen, and hide the content-citation badge/meta in the
tracking detail view, since qwen's source URLs are not reliable enough
to map to brand articles.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Scheduler now runs a token-protected HTTP server on scheduler.http_port
(default 8081, -1 disables) that exposes Prometheus /metrics and the
daily-task JSON snapshot endpoint. Workers gain a blocking Run(ctx) and
are supervised by a WaitGroup: on SIGINT/SIGTERM the metrics server
shuts down first, worker ticks stop scheduling but ongoing runOnce
calls keep their own context and finish, and the process waits up to
60s before exiting. Adds scheduler.http_host/http_port/internal_metrics_token
to config with SCHEDULER_* env overrides and exposes port 8081 in the
Docker image.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds a MonitoringDailyTaskWorker that materializes and dispatches
daily collect tasks from active subscription/desktop-client state,
with per-run atomic metrics and a Prometheus collector. Dashboard
and question-detail APIs now surface a platform_authorization_status
so the UI can distinguish no-desktop-client, no-authorized-platforms,
and authorized states; quota/access-state lookups resolve by request
workspace instead of tenant.
Hardens heartbeat: a desktop_client_primary_leases table gives
per-(tenant, workspace) sticky primary selection, read-mostly lease
resolution avoids a per-heartbeat transaction, and unchanged platform
access reports skip snapshot upserts outside a 10-minute refresh
window. Adds heartbeat primary/snapshot metrics exposed via
token-protected tenant-api /api/internal/metrics endpoints plus
Prometheus. Monitoring quota seed is removed from dev-seed since
daily plans now derive from desktop bindings, and the projection
uses the canonical six-platform catalog so platforms with zero
samples still render.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Introduces current-workspace context helpers plus a WorkspaceScope
middleware that validates X-Workspace-ID against workspace memberships,
defaulting to the actor's primary workspace. Prepares monitoring paths
to resolve quota/access-state per request workspace instead of per tenant.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- desktop: new DeepSeek page adapter and monitor registry
- desktop: extract generic AI auth helpers into shared module and
skip binding when challenge signals are present
- admin-web: render DeepSeek HTML answers with inline citation anchors
and surface reference-number badges on source cards
- server: fall back to inline_links/source_panel_links and read
text/siteName fields when extracting DeepSeek citations
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- drop brand_time_buckets and sparkline/coverage UI; dashboard now returns a single day
- compute mention rate from total actual samples instead of averaging per-platform rates
- canonicalize ai_platform_id across services and aggregate platform-overview snapshots
- strip mojibake and non-answer noise from callback payloads and question detail answers
- keep Kimi citations panel-only; still store raw search_results for audit
- render linked [n] inline citations for qwen/yuanbao answers with scroll-to-source behavior
- restrict立即采集 to today's business date with a clear i18n hint
- refresh content-citations into an ant-design table with truncation and hover styling
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
root