Stable multi-source ingestion
Google / Jina / Firecrawl search, RSSHub with two-layer failover, IMAP email, inbound webhooks, Google Drive, and scholarly databases — every collector ships with health checks and retry policies.
Product · Architecture & capabilities
A production-grade content aggregation stack.
OctoReport is built for real workloads, not demo dashboards — multi-source ingestion, AI cleaning, versioned content, and atomic billing, each engineered for data consistency, observability, and fault tolerance.
No credit card · 10,000 free credits
Design principles
Three principles run through the whole pipeline.
Ingestion, content, billing — every stage is built to production standards, not demo standards.
A versioned content layer
URL-level deduplication: a successful re-collection supersedes the old version, a failure only updates the timestamp. Reports read valid versions only, and every change is traceable.
Auditable atomic billing
Credit deduction runs in transactions with row locks and automatic compensation. Every LLM call gets its own log; every balance change leaves a transaction record.
Modules
Nine modules, each with one job.
Every stage of the pipeline owns its responsibility — and its observability.
Multi-source aggregation
Google · Jina · Firecrawl · RSS · email · webhooks · Google Drive · scholarly APIs · tech communities, all under one collector spec.
- URL dedup
- Auto retry
- Failure alerts
RSSHub hybrid architecture
Embedded and external instances switch automatically by priority. Relative paths like /rsshub/github/trending work out of the box.
- Instance monitoring
- Seamless failover
- Full logs
LLM cleaning
Your custom prompt plus a choice of models extracts titles, summaries, and key fields while filtering ads and navigation noise.
- Billed separately
- Failure isolated
- Dual-format output
Version management
Duplicates are detected by normalized URL: a successful re-collection expires the old version, and reports read unexpired content only.
- Indexed
- Configurable policy
- Version traceability
Report orchestration
Drag-and-drop step configuration with variable injection ({library:N} {source:N} {step:N-1}); each step feeds the next.
- Ordered execution
- Cost tracking
- Per-step retry
Observability
BullMQ queues process asynchronously with RETRYING and FAILED as distinct states, per-step timing, and instantly visible failure reasons.
- Queue priorities
- Redacted logs
- Manual retry
Multi-tenant isolation
Every query is filtered by userId, API keys are stored with AES-256-GCM encryption, and routes are permission-guarded.
- userId scoping
- Route guards
- Audit logs
Transparent billing
Per-call or per-token pricing inside atomic transactions — failed charges roll back. Every cost is traceable, and balance pre-checks prevent overspend.
- Row locks
- Auto-compensation
- Balance pre-check
Flexible scheduling
Interval, weekly, or cron. Failed tasks retry up to 3 times with exponential backoff, with email alerts on failure or low balance.
- Cron support
- Timezone-correct
- Full history
When a collector fails, it should be able to explain why — not silently leave your morning report blank.
Tech stack
Five layers, one operations surface.
From UI to storage, from collection to delivery — every layer is explicit, and every stage is observable, retriable, and reversible.
- Frontend
- Next.js 15 · React 19 · Tailwind 4 · shadcn/ui
- Application
- Custom auth (email OTP) · Prisma 6 · Zod · Vercel AI SDK 5
- Orchestration
- BullMQ 5 · Redis 7 · 13 queues splitting the workload
- Storage
- PostgreSQL 15 · Cloudflare R2 · 12 key indexes
- Operations
- PM2 · Sentry · Pino · primary/backup/DR across three servers
Screenshots
This is the real UI, not a concept render.
Actual captures from the console and share pages, taken July 2026.
Give it 30 minutes and slot it into your team’s morning routine.
Sign up and run on real data: get your first collector live and your first report in your inbox — not a demo, but the report you will actually read next Monday at 8 a.m.
10,000 free credits · No credit card · Export your data anytime