SDK · npm install @agentlog/sdk · MIT license

The Control Layer
for autonomous agents.

Connect every agent running across Zapier, n8n, Make, Claude, and OpenAI to a single governance gateway. Score behavioral risk, halt runaway token loops, and generate legal-grade SOC 2 packages that pass enterprise procurement — in 10 minutes.

Active circuit breaker Legal-grade SOC 2 exports Free tier forever
AgentLog Your Company
RISK 62
LIVE
Events
Compliance
Agents
14:32:07
bulk_update · Zapier · 4,847 rows
CRIT
14:31:58
data_export · n8n · emails.csv
HIGH
14:31:44
send_email · Make · 312 recip.
MED
14:31:12
record_read · Salesforce · PII
HIGH
Auditflow · SOC 2
82% CC8 GAP

Monitors AI agents running across your entire stack

Zapier n8n Make Claude API OpenAI HubSpot Salesforce Airtable Notion Slack Stripe
The platform

One control gateway. Every platform your agents touch.

OpenAI builds tools for OpenAI. Microsoft builds tools for Azure. Neither can govern an enterprise running Claude + n8n + Zapier simultaneously. AgentLog sits above the model layer as the independent cross-platform control authority.

01 · INGESTION

Cross-platform governance gateway

One SDK. Every agent across Zapier, n8n, Make, Claude, and OpenAI feeds a single tamper-evident control plane. Cryptographic hashes. Immutable blocks. When your auditor asks — they get everything, from every platform, in one place.

timestamp 2026-05-15T14:32:07Z
platform zapier
action_type bulk_update
record_count 4847
risk_level CRITICAL
risk_flags [bulk_operation, unusual_hour, pii_touched]
02 · INTELLIGENCE

Active circuit breaker

14-day behavioral baselines per agent. When risk crosses threshold, the circuit breaker fires — halting the operation before it completes. A bulk update at 2AM touching PII scores 96. The breaker trips. The damage doesn't happen.

bulk_operation35
+ unusual_hour62
+ pii_touched87
CIRCUIT BREAKER · operation halted
03 · COMPLIANCE

Legal-grade SOC 2 exports

Claude API maps your control log to Trust Service Criteria and writes plain-English control narratives. Not a dev tool. A compliance workflow. The document your auditor accepts the same day your enterprise prospect asks for it.

SOC 2 EXPORT · MAY 2026
CC6.1 Logical access✓ Documented
CC7.2 System monitoring✓ Documented
CC8.1 Change management✓ Documented
A1.1 Availability✓ Documented
04 · ALERTING

Instant alerts + weekly Control Layer digest

Critical control violations fire instant email and SMS. Every Monday, a plain-English summary of what your agents did, what the circuit breaker caught, and what to fix before the next audit — the retention hook that keeps compliance officers engaged forever.

Circuit breaker fired: bulk_update
Control Layer halted zapier-crm-sync at 4,847 records — 94% outside normal window. Operation blocked before completion.
CRITICAL · HALTED Just now
3 lines
SDK install
8+
Platforms monitored
$299/mo
Target ACV Pro plan
114×
LTV:CAC ratio
EU AI Act · High-Risk Systems
79 DAYS REMAINING

August 2, 2026.
EU AI Act enforcement begins.

High-risk AI systems must demonstrate documented human oversight, continuous audit trails, and behavioral monitoring — or face fines up to 3% of global annual turnover. Your Control Layer is the compliance infrastructure. It deploys in 10 minutes.

Required · Art. 9
Risk management system
Continuous behavioral monitoring per agent, with documented risk thresholds and escalation paths
Required · Art. 12
Logging & traceability
Immutable tamper-evident audit trail for every agent action across all connected platforms
Required · Art. 14
Human oversight
Active circuit breaker with documented override controls and human-in-the-loop escalation
Time to enforcement
79
Days
:
00
Hrs
:
00
Min
AgentLog covers
Article 9 risk management
Article 12 logging & traceability
Article 14 human oversight
Article 13 transparency docs
How it works

Deploy the Control Layer in 10 minutes.

01

Install the SDK

3 lines of code in your app, or drop a webhook URL into Zapier/n8n/Make. No changes to existing automations.

02

Control events start flowing

Every agent action appears in your dashboard with platform, action type, data category, and real-time risk score.

03

Circuit breaker activates

Behavioral baselines and the correlation engine produce a living risk score. Trips the breaker before damage is done — not after.

04

Export for your auditor

One click generates a SOC 2 narrative, EU AI Act documentation package, or GDPR Article 30 record. Claude API writes it in plain English.

SDK · 3 lines

The install that deploys your Control Layer.

Drop this into any JavaScript project. The SDK wraps your agent calls in a control layer that sends structured events to AgentLog in the background — no latency, no code changes to your logic.

javascript · npm install @agentlog/sdk
import { AgentLog } from '@agentlog/sdk';

const log = new AgentLog({ apiKey: 'al_live_...' });

// Wrap any agent action — that's it.
await log.track({
  agentId:     'zapier-crm-sync',
  platform:    'zapier',
  actionType:  'bulk_update',
  dataCategory:'crm_contacts',
  recordCount: 847,
  outcome:     'success',
});
Get started

Your agents are running right now.
Is the Control Layer watching them?

Free tier forever. Pro plan at $299/month. EU AI Act compliant out of the box.

Deploy the Control Layer free See pricing

FREE · 500 EVENTS/MO · NO CREDIT CARD · EU AI ACT READY