Rival Radar

Automated competitive intelligence for B2B SaaS teams.

Track competitor websites, pricing pages, and blogs. Get a Claude-generated Slack brief every week summarising what changed and what it signals — no manual checking, no noise.

Live demo: https://rival-radar-247626835860.us-central1.run.app
API docs: https://rival-radar-247626835860.us-central1.run.app/docs

---

What you get

Add a competitor, hit Run Now, and in ~60 seconds you get a brief like this in Slack:

🎯 Rival Radar — Weekly Brief | August 18, 2026

HubSpot — Pricing page title tag updated from "CRM & Sales Software" to "Marketing Software Pricing" — a shift away from all-in-one framing toward marketing-specific buyers. Suggests possible unbundling of their platform narrative.

Recommended action: Update battlecards for AEs in competitive deals. De-emphasise CRM comparison; lead with marketing workflow depth.

The analyst node reads the diff, interprets what it signals, and the writer turns it into something your sales team can actually use.

---

How it works

Cron trigger (Monday 09:00) or manual Run Now
              │
              ▼
     ┌─────────────────────────────────────────┐
     │           LangGraph pipeline             │
     │                                          │
     │  scraper ──► analyst ──► writer ──► notifier
     │                                          │
     └─────────────────────────────────────────┘
          │              │              │
    async HTTP      Claude Sonnet    Slack
    + SHA-256       interprets       webhook
    diff vs DB      the signal

Node	What it does
scraper	Async-fetches each URL with aiohttp, strips HTML, SHA-256 diffs against the last snapshot. Batch DB prefetch — 2 queries per run regardless of competitor count.
analyst	Feeds each changed excerpt to Claude with competitive-intelligence prompting. Returns structured DiffEntry objects.
writer	Claude synthesises all diffs into a concise, actionable Slack brief.
notifier	Posts the brief to your Slack channel via webhook.

---

Features

• Multi-user SaaS — email/password auth + Google OAuth, session cookies (HttpOnly, SameSite=Lax, Secure), per-user data isolation
• SSRF protection — validate_url_safe() blocks private IPs, link-local ranges, and reserved hostnames before any outbound fetch
• Concurrent scheduling — ThreadPoolExecutor(max_workers=4) runs competitors in parallel; one failure doesn't block others
• Observability — full Langfuse tracing on every LLM call, run history with timestamps and status in the dashboard
• Rate limiting — slowapi on auth endpoints, keyed by real client IP behind Cloud Run's load balancer
• CI/CD — GitHub Actions: ruff → mypy → pytest → Docker build → GCP Cloud Run deploy (deploy blocked until tests pass)

---

Stack

Layer	Tech
Agent framework	LangGraph
LLM	Claude Sonnet 4 via langchain-anthropic
API	FastAPI + uvicorn
Auth	itsdangerous sessions, bcrypt, Google OAuth 2.0
Database	SQLAlchemy 2.0 — SQLite (dev) / Postgres (prod)
Scheduler	APScheduler — weekly cron, Monday 09:00
Observability	Langfuse
CI/CD	GitHub Actions → GCP Artifact Registry → Cloud Run
Infra	Docker, Terraform-free (single Cloud Run service)

---

Quickstart

git clone https://github.com/Akhilvallala1/rival-radar
cd rival-radar
cp .env.example .env          # fill in ANTHROPIC_API_KEY + SLACK_WEBHOOK_URL
pip install -e ".[dev]"

Scrape a URL and see the diff:

python -m rival_radar scrape --url https://competitor.com/pricing --name "Acme Corp"

Run the full AI pipeline:

python -m rival_radar run --competitor-id 1

Start the web dashboard:

uvicorn rival_radar.api:app --reload
# open http://localhost:8000

---

API

# Health check
curl http://localhost:8000/health

# Add a competitor
curl -X POST http://localhost:8000/competitors \
  -H "X-API-Key: changeme" \
  -H "Content-Type: application/json" \
  -d '{"name":"HubSpot","urls":["https://hubspot.com/pricing"],"cadence":"weekly"}'

# Trigger a run immediately
curl -X POST http://localhost:8000/competitors/1/run \
  -H "X-API-Key: changeme"

# View recent briefs
curl http://localhost:8000/runs -H "X-API-Key: changeme"

Interactive docs at /docs (Swagger UI).

---

Development

ruff check src/ tests/    # lint
mypy src/                 # type check
pytest tests/ -v          # 95 tests across auth, multi-tenancy, SSRF, scheduler, integration

All three gates run on every PR via GitHub Actions before any deploy.

---

Deployment

Every push to master builds a Docker image, pushes to GCP Artifact Registry, and deploys to Cloud Run — but only after CI passes.

Required GitHub secrets:

Secret	Description
GCP_PROJECT_ID	GCP project ID
GCP_CREDENTIALS	Service account key JSON
ANTHROPIC_API_KEY	Anthropic API key
SLACK_WEBHOOK_URL	Slack incoming webhook
DATABASE_URL	Postgres connection string
SECRET_KEY	Session signing key (random 32+ char string)
LANGFUSE_PUBLIC_KEY	Langfuse public key (optional)
LANGFUSE_SECRET_KEY	Langfuse secret key (optional)

---

Roadmap

• Tier enforcement — starter (2 competitors, weekly), pro (10, daily), team (unlimited, hourly); enforced server-side on every request
• G2 review monitoring — structured extraction of rating, review count, and recent reviews via JSON-LD + CSS fallback; hashes structured data so cosmetic G2 page changes don't trigger false positives
• Keyword alerts — instant Slack ping when a tracked keyword appears in a scrape diff; API: POST/GET /competitors/{id}/alerts, DELETE /alerts/{id}

---

License

Business Source License 1.1 — free to self-host for internal use. Commercial hosting or reselling requires a separate license. Converts to MIT on 2029-01-01.

Contact akhilvallala0115@gmail.com for commercial licensing.