Sealed Pipeline

Sealed Pipeline

The sealed pipeline is the parent civilization’s nervous system. Every build event, every telemetry batch, every finding flows through it.

The 7 Stages

ORCHESTRATE → EMIT → ACCUMULATE → ROLLUP → INSCRIBE → RECALL → BUILD

Stage	What Happens	Frequency
ORCHESTRATE	Decide what needs monitoring	Continuous
EMIT	Capture raw events (build, telemetry, findings)	Per-event
ACCUMULATE	30-second cycle summaries in DO SQLite	Every 30s
ROLLUP	Hourly/nightly synthesis by Joe	Hourly
INSCRIBE	Write findings to COG	Per-rollup
RECALL	Load priorities into next session	Per-session
BUILD	Monitoring priorities inform next build cycle	Per-build

Monitoring the Pipeline

# Check pipeline health
curl -s "$API/api/monitoring/rollup/priorities" -H "Authorization: Bearer $AUTH_SECRET"
# Returns: { priorities: [...], health: 1.0, generated_at: "..." }

• health: 1.0 — pipeline is flowing normally
• health < 0.8 — investigate ACCUMULATE or ROLLUP stages
• priorities: [] — no issues flagged (clean)
• priorities: […] — items need attention

Pipeline for Customer Orgs

When a customer’s probe sends telemetry, it enters the same pipeline:

1. EMIT — probe telemetry event captured
2. ACCUMULATE — 30-second summary in the monitoring agent’s DO SQLite
3. ROLLUP — hourly synthesis produces governance score update
4. INSCRIBE — findings written to govern_findings
5. RECALL — customer dashboard loads fresh data

The parent and child share the pipeline infrastructure, but the data is isolated by org_id at every stage.