Starter
500 requests / day
Up to 30 requests per minute
Five active keys · scoped permissions · usage dashboard
Request StarterAEGLIX MIND / DEVELOPER PLATFORM v1
Analyze suspicious URLs, messages and screenshot text. Connect public-source evidence and vulnerability context to your own applications.
Base URL: https://aeglixmind.com/api/v1/integrations. Call from your server. Never embed a key in frontend JavaScript, a mobile app bundle, a URL or a public repository.
curl https://aeglixmind.com/api/v1/integrations/guardian/analyze \
-H "Authorization: Bearer $AEGLIX_API_KEY" \
-H 'Content-Type: application/json' \
--data '{"kind":"url","url":"example.com","consent":true}'| Operation | Permission | Returns |
|---|---|---|
POST /guardian/analyze | guardian:write | URL, message or screenshot text assessment |
GET /guardian/analyses/{id} | guardian:read | An unexpired assessment created by that key |
POST /intelligence/lookup | intelligence:read | Bounded public feed evidence and entity IDs |
GET /entities/{id}/relationships | relationships:read | Public feed relationships, sources and retrieval dates |
GET /cves/{cve_id} | cves:read | Indexed NVD vulnerability and CISA context |
Suffixes above follow the base URL. Requests and responses use JSON. Campaigns and Arsenal are not included in this API.
Choose exactly one kind and content field. consent must be the JSON boolean true.
{"kind":"url","url":"https://example.com/path","deep_scan":false,"consent":true}
{"kind":"message","text":"An unexpected message asking for a money transfer.","consent":true}
{"kind":"screenshot","image_base64":"BASE64_IMAGE_BYTES","consent":true}Messages: 10–12,000 characters. Screenshots: still PNG/JPEG/WebP up to 4 MiB and 12 megapixels, standard base64 without a data URI prefix. Total JSON body: at most 6 MiB. deep_scan:true opts into bounded URL content fetching. URLs in message or screenshot text are not separately scanned.
A completed analysis uses schema_version: guardian-assessment-v1 and returns id, kind, status, created_at, expires_at and result. URL results include risk level, factors, explanation, ML/model analysis and unavailable providers. Message results include verdict, summary, warning signs, recommended actions and limitations. Score and confidence are not safety guarantees.
Lookup body: {"kind":"domain","value":"example.com"}. Kinds: domain, url, ip, hash and cve. Hashes use sha256:HEX, sha1:HEX or md5:HEX. An empty result means unknown, not safe. Public relationships are source-reported, not independent confirmations or actor attribution. Indexed CVE reads do not fetch new NVD records on demand.
Provider coverage can be incomplete. A URL analysis may complete with unavailable providers or model augmentation; inspect those fields before taking action.
const response = await fetch(
'https://aeglixmind.com/api/v1/integrations/guardian/analyze', {
method: 'POST',
headers: {
Authorization: `Bearer ${process.env.AEGLIX_API_KEY}`,
'Content-Type': 'application/json'
},
body: JSON.stringify({kind: 'message', text: message, consent: true}),
signal: AbortSignal.timeout(185000)
});
if (!response.ok) throw new Error(`AEGLIX API: ${response.status}`);
const assessment = await response.json();import os
import requests
response = requests.post(
'https://aeglixmind.com/api/v1/integrations/guardian/analyze',
headers={'Authorization': f'Bearer {os.environ["AEGLIX_API_KEY"]}'},
json={'kind': 'url', 'url': 'example.com', 'consent': True},
timeout=(10, 185),
)
response.raise_for_status()
assessment = response.json()Use GET /guardian/analyses/{id} with the same key to retrieve a completed result. Store assessment IDs in your system; identifiers do not grant access without the creating key.
Daily allowance is volume, not simultaneous processing capacity. Analysis runs synchronously, with a 180-second application deadline and bounded shared capacity. Start with one or two requests in flight and increase only after observing latency and capacity responses. Higher daily allowances do not promise reserved concurrency or a response-time SLA.
Admitted responses expose X-RateLimit-Limit, X-RateLimit-Remaining, X-RateLimit-Reset (UTC epoch seconds), and X-RateLimit-Key-Remaining. Your per-key cap can be lower than the shared plan allowance.
| Status | What to do |
|---|---|
| 401 | Check the key, expiry and revocation status. |
| 403 | Check the required permission and active premium plan. |
| 404 | Record unavailable, expired or inaccessible to this key. It is not a clean verdict. |
| 413 / 415 / 422 | Fix input size, JSON content type, consent or field values before retrying. |
| 429 | Daily or per-minute allowance reached. Respect Retry-After. |
| 503 | Capacity or provider unavailable. Back off using Retry-After with jitter. |
Errors include a stable error.code, a readable message and a support request_id, also returned as X-Request-ID. Keep the request ID for support; never log key secrets or private submissions.
There is no queue, bulk endpoint, webhook or idempotency guarantee in this release. A timed-out or retried POST may consume another request and create another assessment. Do not retry blindly. Client or reverse proxy timeouts can be shorter than the application deadline.
Premium API access. Contact us for pricing and activation; no payment is taken on this page.
Starter
Up to 30 requests per minute
Five active keys · scoped permissions · usage dashboard
Request StarterGrowth
Up to 60 requests per minute
Five active keys · scoped permissions · usage dashboard
Request GrowthScale
Up to 120 requests per minute
Five active keys · scoped permissions · usage dashboard
Request ScaleAllowances are shared across every key in your account and reset at 00:00 UTC. Creating or rotating keys does not add requests. Need a larger allowance? Discuss custom capacity.
One admitted API call uses one request, including result reads, invalid submissions, missing records and upstream failures. Authentication failures, denied scopes and rejected quota/rate admissions do not use requests. Key management and viewing console usage do not consume your integration allowance. No automatic overage charges.
Only submit content you are authorized to process. Consent permits the configured external threat/model services to process it. Redact personal data and credentials first. Screenshot OCR runs locally, then sends extracted text for model analysis immediately; use reviewed message text if you need a redaction step.
Assessment retrieval expires after seven days and is restricted to the creating key. Revoking a key also prevents retrieval of its prior assessments; export needed results before a planned rotation. URL scans also enter the account’s Guardian history under its separate retention rules. Original messages, images and OCR text are not saved separately, but generated results may contain excerpts. See the privacy policy and terms of use.
Need help activating or changing an allowance? Contact API support with your account email and key ID, never the key secret or unredacted submissions.