Map every page, endpoint, and auth flow in a target web app.
SiteAtlas runs browser-automation profiles against any web app and produces a structured inventory: pages, API endpoints, auth flows, forms, dead links, and screenshots. Built for AI-agent discovery, M&A technical diligence, and compliance audits.
pnpm install pnpm dev > Resolving run profile: standard_mapping
> Allocating runner from pool...
> Launching Playwright (stealth)
> Capturing 3 modes: nav + screenshots + DevTools
> Discovered 247 pages, 89 endpoints, 14 auth flows
> Saved artifacts to R2 (signed URLs, 24h TTL)
> Generated API docs: api-docs.json + api-docs.md
> Bundle ready: bundle_zip (8.3 MB)
✓ Run 8d3f9a21 complete
pages: 247
endpoints: 89
auth: 14
cost: $0.42
duration: 4m 12s Four modes, one engine.
Pick the profile that matches the target environment. Politeness defaults are enforced — never use deep_mapping on production without operator approval.
standard_mapping
standardDefault. Balanced coverage. Production-friendly politeness.
deep_mapping
deepExhaustive. Discovers hidden endpoints, auth flows, edge cases.
demo_mapping
smokePolite. Light footprint. For demos and proof-of-concept.
api_inventory_only
smokeEndpoints only. No UI navigation. Fastest.
Six steps from URL to inventory.
Resolve
Look up the workspace, project, and run profile. Apply target budget preset (smoke / standard / deep).
Allocate
Pull a runner from the browser pool. Stealth Playwright launch. Polite defaults (rate limit, robots).
Navigate
Three-mode capture: page navigation, screenshots, DevTools payload + response. Click through forms, follow links.
Discover
Extract pages, API endpoints, auth flows, forms. Auto-classify (login, signup, checkout, dashboard, etc).
Artifact
Bundle: api-docs.json + api-docs.md, screenshots, raw responses. Store in R2 with signed URLs (24h TTL).
Report
Run record in Convex. Status visible in the local operations console. Share link with TTL for read-only access.
Production tools, no driver code.
Next.js + Convex
Control plane shell, multi-tenant schema, workspace + member + API-key management, billing policy with plan limits.
Playwright runner
Three-mode capture: navigation, screenshots, DevTools payload/response. Profile-driven.
Upstash managed queue + ledger
Production queue provider (option), shared run state across regions. Local-only queue as the dev baseline.
S3-compatible artifacts (CF R2)
Signed URL delivery, local repro bundle (bundle_zip) for debugging, dead-letter tracking for max-attempt failures.
Cron + run schedules + templates
Project run schedules with manual/cron trigger support. Reusable run config presets. /api/schedules/tick endpoint.
Local operations console
Run filtering, queue controls, lifecycle actions, cleanup. Expiring share links for read-only run reports.
Self-host, or use the managed control plane.
The runner and Convex schema are MIT. The managed control plane handles queue, ledger, and R2 artifacts for you.
Pro
- 50,000 runs / month
- 5 workspaces
- All run profiles
- Convex managed backend
- Priority email support
Team
- 500,000 runs / month
- 25 workspaces
- Run templates + schedules
- Audit log export
- Dedicated Slack channel
Open source. Self-hostable. Or let us run it.
Read the source. Run the runner. Or sign up for the managed control plane.