Ask anything. Choose sources when needed.
One composer handles quick web search, selective source search, and deep research. Clarifying questions appear in the run chat when the agent needs direction.
Session
Local workbench
Live API context
Evidence
Fresh + selective
Web, profiles, source sets
Agents
Surface to report
Plan, critique, trust
Web search
Snippet or extracted content across federated providers with semantic cache.
01
Query
Fresh web candidates
02
Extract
Readable source content
03
Rank
Recency + source quality
Selective search
Federated across curated source sets with SSRF-guarded runtime and per-source rankers.
Profile
Curated scope
Source profiles and sets
Guard
SSRF checked
Safe connector runtime
Rank
Fast to quality
Mode-aware result ordering
Intelligence alerts
Scheduled monitors watch web, selective, or research workflows and collect new events into in-app and email digests.
Active
0
Enabled monitors
New
0
Unacknowledged events
Failed
0
Checks needing attention
Next
-
Upcoming run
Digest preview
select a monitor to preview its digest
Saved library
Collect search results, assets, research runs, and monitor events with annotations for reuse and exports.
Collections 0 Reusable folders
Exports CSV / PDF Team-ready handoff
Tags Taxonomy Auto-classified signals
Battlecards
Turn saved intelligence and monitor signals into competitor-facing summaries for sales, product, and strategy teams.
Integrations
Configure delivery channels and signed outbound webhooks for Silo, CRMs, team chat, and mobile alerts.
Source profiles
Named source bundles saved to sources.yml. Use source_profile: "name" in selective search requests.
Add source / max 10 per profile
Save to sources.yml
Source sets and credentials
API service
Create tenant API keys, watch request volume, and inspect recent calls across search, assets, research, monitors, and connectors.
Requests
0
Recent tracked calls
Units
0
Metered usage
Errors
0
4xx and 5xx
Latency
0ms
Average response time
Use Croox API keys
Authorization: Bearer crx_YOUR_KEY
curl
Fetch an image asset
curl -X POST http://127.0.0.1:8001/v1/assets/search \
-H "Authorization: Bearer crx_YOUR_KEY" \
-H "Content-Type: application/json" \
-d "{\"query\":\"Barack Obama\",\"type\":\"image\",\"limit\":1}"
python
Run web search
import httpx
headers = {"Authorization": "Bearer crx_YOUR_KEY"}
body = {"query": "latest AI research tools", "depth": "content"}
response = httpx.post("http://127.0.0.1:8001/v1/search", headers=headers, json=body)
print(response.json())
Console session
Sign in with OIDC
Log out
Refresh session
Deep research
Multi-agent planner -> subagents -> writer -> critic. Stream live, interject mid-run, render md/html/pdf.
Surface
Clarify direction
Evidence-informed choices
Plan
Decompose work
Waves and subagents
Critic
Audit claims
Revision gate
Trust
Score evidence
Citations and source quality
...
pending tier - mode - cost $0 waves 0 subs 0 wall 0s
Clarify research direction
Execution granularity
single run with subagents
child runs with parent synthesis
Create final plan
Trace 0 events
Load stored trace
Report
Download .md
Open .html
View JSON
Download .pdf
