Skip to main content

Transmission

Transmission ensures that reliability signals — detections, alerts, incident events, recovery state — move correctly between components of the ACME stack. It’s the connective tissue that keeps your reliability infrastructure coherent.

Why Signal Integrity Matters

Each component in the ACME stack produces signals: Sentinel detects a stall, Watchdog misses a heartbeat, OCTriage classifies an incident. For your reliability infrastructure to work as a whole, those signals need to:
  • Arrive — reach the right destination
  • Arrive intact — not corrupted or truncated in transit
  • Arrive in order — timelines must be reconstructible
  • Arrive once — no duplicated events inflating incident counts
Without this, your tools disagree. Agent911 shows one state; Sentinel shows another. You spend incident time debugging your monitoring, not your agents. Transmission solves this.

What Transmission Does

Event Delivery Guarantees

At-least-once delivery with deduplication. Events arrive, or you know they didn’t.

Order Preservation

Detection events, classification results, and recovery state maintained in sequence.

Coherence Verification

Validates that signal consumers (Agent911, alert channels) received what was sent.

Dead Letter Handling

Failed deliveries captured and surfaced for operator review — nothing silently dropped.

Signal Architecture

Transmission operates as the transport layer between ACME components:
┌──────────────┐    ┌────────────────────┐    ┌──────────────────┐
│  Sentinel    │───▶│                    │───▶│    Agent911      │
│  Watchdog    │───▶│   Transmission     │───▶│    Alert Channel │
│  DriftGuard  │───▶│   Signal Router    │───▶│    Webhooks      │
│  SphinxGate  │───▶│                    │───▶│    Operator CLI  │
└──────────────┘    └────────────────────┘    └──────────────────┘

                    ┌────────▼────────┐
                    │  Event Journal  │
                    │  (append-only)  │
                    └─────────────────┘
The event journal is the source of truth. Every signal from every component is written there before delivery. This enables:
  • Replay — reconstruct exactly what happened and when
  • Audit — prove what signals were sent and received
  • Recovery — resume delivery after a disruption without losing events

Signal Types

Transmission handles four categories of reliability signal:
CategoryExamples
Detection eventsStall detected, heartbeat missed, drift alert, routing anomaly
Classification resultsOCTriage incident classification, confidence score, evidence bundle path
State transitionsAgent health change, recovery started, recovery verified
Operator actionsSnapshot triggered, playbook started, bundle exported
Each signal type has a defined schema, ensuring consumers know exactly what they’re receiving.

Delivery Channels

Transmission routes signals to configured destinations:
# ~/.acme/transmission.yaml
channels:
  agent911:
    enabled: true
    endpoint: local            # Local Agent911 instance

  webhooks:
    - name: "pagerduty"
      url: https://events.pagerduty.com/v2/enqueue
      secret: ${PAGERDUTY_KEY}
      filter:
        severity: [HIGH, CRITICAL]   # Only escalate high-severity events

    - name: "slack-alerts"
      url: https://hooks.slack.com/services/...
      filter:
        types: [STALL, CRASH, RECOVERY_FAILED]

  email:
    enabled: true
    to: oncall@yourcompany.com
    filter:
      severity: [CRITICAL]

Event Journal

Every event is appended to the Transmission journal before delivery:
/var/acme/transmission/journal/
  events-2026-03-09.jsonl
  events-2026-03-10.jsonl
  ...
Journal format:
{
  "event_id": "evt_9xk2m4",
  "timestamp": "2026-03-09T15:42:31Z",
  "source": "sentinel",
  "type": "STALL_DETECTED",
  "agent": "support-agent-prod",
  "severity": "HIGH",
  "payload": { "duration_seconds": 262, "provider": "openai" },
  "delivery": {
    "agent911": { "status": "delivered", "at": "2026-03-09T15:42:31.08Z" },
    "webhooks": [
      { "name": "pagerduty", "status": "delivered", "at": "2026-03-09T15:42:31.12Z" }
    ]
  }
}

Dead Letters

If a signal fails to deliver after retries, it’s placed in the dead letter queue — not silently dropped:
# View dead letter queue
acme transmission dead-letters

# Retry failed deliveries
acme transmission retry --channel webhooks

# Inspect a specific failed event
acme transmission inspect evt_9xk2m4
Dead letters are surfaced in the CLI and optionally in Agent911.

Quick Start

# Check Transmission status
acme transmission status

# View event journal (recent)
acme transmission log --tail 20

# Verify delivery for a specific agent
acme transmission verify --agent my-agent-name

# Test channel delivery
acme transmission test --channel pagerduty

CLI Reference

# Status
acme transmission status
acme transmission status --format json

# Event log
acme transmission log
acme transmission log --tail 50 --follow
acme transmission log --since 1h
acme transmission log --type STALL_DETECTED --format json

# Journal
acme transmission journal list
acme transmission journal export --date 2026-03-09 --output events.jsonl

# Delivery management
acme transmission verify --agent <name>
acme transmission dead-letters
acme transmission retry --all
acme transmission test --channel <name>

# Configuration
acme transmission config show
acme transmission config validate

Integration

Transmission doesn’t require configuration if you’re only using ACME tools locally — it operates as an internal transport layer by default. You configure it explicitly when:
  • Connecting to external channels (PagerDuty, Slack, email, custom webhooks)
  • Running multi-machine deployments where components are on different hosts
  • Enabling compliance logging that needs verifiable delivery receipts
  • Replaying events after a disruption or missed alert window

Pricing

Transmission is a paid product, included with the Operator Bundle or available standalone. See Pricing for current rates.

Next Steps