Try a search — Zagros

UI preview · static fixture data

Preview the search interface.

This page uses 8 hand-picked fixture records to preview the UI only. It does not call HelixDB, does not run Zagros BM25, and does not show runtime scores. Use Try Zagros Live below for the real Rust retrieval engine and seeded database.

Fixture preview only. Matching here is a simple browser-side text filter. Real ranking, thresholds, field weights, and exact/phrase bonuses exist only in the Rust Zagros runtime.

Live demo · GitHub Codespaces

Run a real Zagros instance.

Launch Zagros in your own GitHub Codespace. The dev container starts HelixDB, MCP, and the real web UI, then seeds 100 CVEs plus CWE, ASVS, CAPEC, and ATT&CK. Port 8788 is forwarded privately to your GitHub account by default.

Try Zagros Live →View Codespaces config →

GitHub Codespaces usage is billed or quota-limited by GitHub according to your account. This launches your own disposable instance; it is not a shared public Zagros server.

MCP preview

What your agent sees.

→

search_cves — real MCP tools/call request

// JSON-RPC body sent to POST /mcp
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "search_cves",
    "arguments": {"query": "xz", "top_k": 5}
  }
}

→

get_cve — real MCP tools/call request

{
  "jsonrpc": "2.0",
  "id": 2,
  "method": "tools/call",
  "params": {
    "name": "get_cve",
    "arguments": {"cve_id": "CVE-2024-3094"}
  }
}

MCP reference →

Example investigation flow

How an agent can use Zagros evidence.

This is a documented example, not live output from this static page. For the question "What does Zagros know about CVE-2024-3094, and what should I do next?", an agent should check current corpus health, retrieve evidence, and keep its own interpretation separate.

Evidence — example of fields to verify

index_status → inspect current overall_health at runtime
get_cve → retrieve current stored CVE record
ID: CVE-2024-3094
Finding: malicious code in upstream xz/liblzma 5.6.0 and 5.6.1 release tarballs
Impact stated by record: could interfere with sshd authentication in affected Linux environments
Source: CVE Program / cvelistV5
Trust tier: authoritative

The text above is an illustrative excerpt. In a real run, use the current get_cve response and its provenance/source URL.

Agent analysis — separate from Zagros output

The record supports a serious software supply-chain compromise affecting xz/liblzma release artifacts.
Zagros does not infer affected versions, exploitability in your environment,
or remediation status unless those facts are present in authoritative evidence.

This interpretation is performed by the consuming agent, not generated by Zagros itself.

Agent recommendation — based on verified evidence

1. Verify whether the affected product/version exists in your environment.
2. Confirm the current vendor advisory and CVE record.
3. Prioritize remediation if exposure and affected-version evidence match.
4. Re-run the investigation after source refresh if evidence is stale.

These actions are an example agent workflow. Zagros supplies retrieval and provenance; it does not generate recommendations, certify compliance, or autonomously remediate systems.