ScivlyResearch intelligence
Console
Back to docs

API reference

A thin API surface for monitors, digests, questions, and delivery hooks.

The public API is intentionally narrow. It exposes the monitor lifecycle, search, digest orchestration, and evidence-aware questions without leaking internal pipeline mechanics.

Auth

Workspace-scoped bearer keys for platform and self-hosted flows.

Webhook-safe

Delivery events carry deterministic ids for retries and replays.

Low ceremony

A small endpoint surface is easier to maintain across product phases.

Create a monitor

curl -X POST https://api.scivly.dev/v1/monitors \
  -H "Authorization: Bearer scivly_test_key" \
  -H "Content-Type: application/json" \
  -d '{
    "workspace_id": "ws_research",
    "name": "agent-benchmarks",
    "filters": {
      "topics": ["agents", "retrieval", "multimodal reasoning"],
      "authors": ["Research Lab A", "Research Lab B"]
    }
  }'

Authentication flow

1

Issue workspace-scoped API keys for backend services and operator tooling.

2

Attach monitor, digest, and question actions to the same workspace context.

3

Replay failed deliveries with idempotency keys instead of one-off scripts.

Endpoint reference

POST/v1/monitors

Create a new paper monitor for a workspace.

GET/v1/papers/search

Search indexed papers with workspace-aware filters.

POST/v1/digests/run

Generate a digest preview or trigger a delivery run.

POST/v1/questions

Ask follow-up questions against retrieved paper context.

Webhook payload

{
  "event": "digest.delivered",
  "workspace_id": "ws_research",
  "digest_id": "digest_2403",
  "status": "sent",
  "paper_count": 12,
  "matched_monitors": ["agent-benchmarks", "vision-reasoning"]
}

Next step

Pair the API with the admin surface.

The API is intentionally small; most operational confidence comes from seeing monitor state, queue health, digest runs, and webhook behavior in one admin console.

Open admin console