Signal Spec

Canonical data model for operational intelligence.

Overview

Signal Spec defines the schemas and types for:

• Signals - Normalized operational observations (tickets, alerts, incidents, findings)
• Root Causes - Persistent clustered issues with lifecycle tracking
• Remediations - Corrective actions with efficacy measurement
• Validation Signals - Evidence of fix effectiveness

Structure

signal-spec/
├── cmd/signal-spec/  # CLI application
│   ├── main.go
│   └── cmd/
│       ├── root.go
│       ├── report.go
│       ├── schema.go
│       └── validate.go
├── pkg/
│   ├── common/       # Shared types (severity, domain, entity)
│   ├── signal/       # Raw signal type
│   ├── rootcause/    # Root cause type
│   ├── remediation/  # Remediation and validation types
│   └── export/       # XLSX report generation
├── schema/           # Generated JSON schemas
├── examples/         # Example payloads
└── docs/             # Architecture documentation

CLI

# Build
go build -o signal-spec ./cmd/signal-spec

# Generate XLSX report from root causes
signal-spec report -i rootcauses.json -o summary.xlsx
signal-spec report -d ./rootcauses/ --leaders leaders.json -o summary.xlsx

# Generate JSON schemas
signal-spec schema generate -o schema/

# Validate a JSON file
signal-spec validate -t signal signal.json
signal-spec validate -t rootcause rootcause.json

Usage

Go Types

import (
    "github.com/plexusone/signal-spec/pkg/signal"
    "github.com/plexusone/signal-spec/pkg/rootcause"
)

// Create a signal
sig := signal.Signal{
    ID:       "sig-001",
    Type:     signal.TypeSupportTicket,
    Severity: common.SeverityHigh,
    Summary:  "OAuth token refresh failures",
    // ...
}

// Create a root cause
rc := rootcause.RootCause{
    ID:     "rc-001",
    Title:  "Redis session replication instability",
    Status: rootcause.StatusActive,
    // ...
}

Core Concepts

Signal → Root Cause Mapping

Signals are raw input. Root causes are interpretations. The mapping is performed by LLM analysis with access to:

• Signal content and history
• Codebase context (via graphize)
• System documentation

Lifecycle States

Root causes progress through states:

NEW → ACTIVE → MITIGATING → VALIDATING → STABLE/REGRESSED → RESOLVED

This enables tracking of:

• Issue persistence
• Remediation effectiveness
• Regression detection
• Operational debt accumulation

Documentation

See docs/architecture.md for detailed design.

License

MIT