henexus Logs

Every log from every system. One viewer.

Vercel, Stripe, AWS, GitHub Actions, Supabase, Sentry — and everything you build yourself — normalized into one searchable stream.

Get Started See the SDK

NormalizedLogEntry

occurred_atserviceleveltitlestatus_codeduration_ms

One schema. Every source.

Filters, alerts, and queries become source-agnostic the moment a log enters henexus.

Vercel JSON

deployment_id: dpl_…
region: iad1
function_name: checkout
duration_ms: 10042

Stripe JSON

event: invoice.payment_failed
customer: cus_K7…
amount: 4900
currency: usd

Supabase webhook

table: orders
type: INSERT
record.id: f3a1…
schema: public

normalize

NormalizedLogEntry

occurred_at: timestamptz
service: string
level: enum
code: string
title: string
transaction_id: string
status_code: int
duration_ms: int
… 17 more indexed columns
structured_fields { … }

25 indexed columns. One JSONB overflow. Zero per-source tables.

15 connectors live on day one.

Plus a Python SDK to write your own.

AWS CloudWatch

pull

Vercel

push

GitHub Actions

hybrid

Cockroach Cloud

pull

Anthropic

pull

OpenAI

pull

LiveKit

push

Supabase

push

Cloudflare Workers

push

Heroku

push

Sentry

push

Stripe

push

Resend

push

Apple App Store

push

TestFlight

hybrid

AllHostingCI/CDDatabasePaymentsEmailError trackingMobileEdgeAIReal-time

An Excel-style log viewer that scales.

Three composable filter layers. Cursor-paginated. Trace-aware.

🔍Search title and description…

SavedMy ErrorsProd 5xxSlow Requests >5sCheckout Flow

FieldsNest: OptiAi ▾After ▾Before ▾Level ▾Service ▾+ Add

Time	Nest	Service	Level	Code	Title	Duration	Trace ID
14:03:17	OptiAi	vercel	warn	503	GET /api/checkout — 503 timeout	10042	0af7651916cd
14:03:15	OptiAi	stripe	error	fail	invoice.payment_failed	—	1c5d3b2a44ef
14:03:12	OptiAi	supabase	info	200	row.insert on orders	8	92aa7c1b5d80
14:03:09	Marketing	resend	info	sent	email.delivered to user@…	—	d2f1e7a9c10b
14:03:07	OptiAi	aws	warn	4xx	ALB target reset	142	3b91e0c44a17
14:03:02	OptiAi	github	info	ok	workflow_run completed	38120	5e8a23ff29bd
14:02:58	Staging	sentry	error	Error	TypeError: Cannot read properties of undefined	—	7c19be01f4a2
14:02:55	OptiAi	cloudflare	info	200	GET /assets/main.js	12	a40d9b2c8e51

428 entries · loaded 50

quota 62%⚙

Three layers of filters

Saved predicates, ad-hoc field chips, and per-column type-to-filter — composable and deterministic.

Cursor pagination

Stable boundaries even when new entries arrive mid-scroll. Same API hen uses.

View correlated

One click on a trace ID filters the table to every related entry across services.

Write a custom connector in an afternoon.

The same Python SDK that powers our 15 built-ins is the one you use. Same shape. Same validation. Same sandbox.

Pydantic v2 validation — unknown fields refused at construction.
Runs sandboxed on EventBridge → Fargate. Egress allow-listed by credential binding.
Same row shape as every built-in. Same viewer. Same filters.
Bundled helpers for HMAC verification, JWT signing, redaction, stable digests.

airflow_connector.py

from henexus_logs import Logger, Log, credential_store
import httpx

class AirflowLog(Log):
    service = "airflow"
    dag_id:  str
    task_id: str
    run_id:  str

def main():
    logger = Logger.factory(connector_id="cn_airflow_prod", project_id="pj_acme")
    since = logger.state.get("since", default="1970-01-01T00:00:00Z")

    with credential_store("airflow-prod") as cred:
        resp = httpx.get(f"{cred.base_url}/api/v1/events",
                         params={"since": since},
                         headers={"Authorization": f"Bearer {cred.token}"})

    for e in resp.json()["events"]:
        log = AirflowLog(title=f"{e['dag_id']} / {e['task_id']} — {e['state']}",
                         level="error" if e["state"]=="failed" else "info",
                         dag_id=e["dag_id"], task_id=e["task_id"], run_id=e["run_id"])
        logger.save_unique(log, unique_by=["dag_id", "task_id", "run_id"])

Save once. Never see again.

Credentials that vanish on save.

Envelope encryption. DEK-per-credential wrapped by tenant CMK in AWS KMS.
No reveal. No copy. No export — ever. Not for the user who saved it. Not for admins.
Lost a key? Rotate. There is no recovery path. Support can't help you read it back.

Before save

Bearer token

↓ after save

After save

Bearer token

••••••••••••••••••••

Rotate

You will not see this again. Rotate to replace.

Failure is loud. Success is silent.

Broken connectors show up in the same viewer you already watch.

logs-selfschema_mismatchvercel14:03:17

Vercel webhook payload changed shape — 247 retries since 13:58

{
  "missing":       ["record.old_record"],
  "unexpected":    ["record.delta"],
  "type_mismatch": [
    { "field": "record.id", "expected": "uuid", "got": "integer" }
  ]
}

quarantine_id: q_8f3d…·retains until 2026-05-08

hen

hen sees what you see — with your permissions.

Opt-in per tenant. Off by default. Logs are high-signal — and high-stakes.
Clustered before hen sees them. Cluster key on (service, level, fingerprint). One summary event per cluster, not per entry.
No standing read. hen queries through your viewer's API with the asking user's permissions — every time.

hen

Error spike on /api/checkout 27 min ago correlates with deploy build #1842 (commit a501f06). The previous 12 deploys to this endpoint had p95 < 800 ms; this one is at 10,042 ms. Want me to draft an alert rule?

Coming next — automated response.

P1 ingestion is shipping with these in mind. Schemas and event paths are ready.

Coming soon

On-call scheduling

PagerDuty-style rotations, schedules, and escalation paths. Pages humans when alerts fire.

Coming soon

Heartbeat & uptime

Active probes of your services. Alert on missed beats or degraded responses.

Coming soon

Custom alert rules

Customer-defined event emitters that watch ingested logs and publish henexusEvents when conditions match.

One platform, one audit trail.

Every connector, credential, filter, and retention change goes to /portal/settings/audit.

Per-Nest, per-environment.

Every entry tagged with environment (prod / stg / tst / dev). Page-header selectors persist across screens.

Three roles, zero ambiguity.

Observer / Developer / Logs Administrator. No plaintext credential ever crosses a permission boundary.

Stop bouncing between dashboards.

Connect your stack in minutes. No credit card required.

Get Started Free