Skip to main content

What are Signals?

Signals is an Overmind feature that builds deployment confidence by analyzing patterns in your infrastructure changes. Instead of simply flagging what's changing and the associated risks, Signals determines whether your changes are actually safe to deploy by examining historical patterns and current context. In essence, it captures and operationalizes all that organizational and change-specific tribal knowledge that typically lives only in your team's collective memory.

When you create a change in Overmind, Signals automatically analyses your infrastructure modifications and provides insights that help teams make informed deployment decisions.

Types of Signals​

In Overmind there are two different types of signals:

Change-Level Signals​

These signals analyse your entire deployment as a whole and appear in the top section of the Signals page. Some examples of change-level signals could be:

  • Routine Analysis: "Recent infrastructure changes span multiple attributes, with modification frequencies ranging from first-time updates to daily deployments"
  • Blast Radius: "200 items" affected by your change (coming soon)
  • Environment Context: "Production" environment detection (coming soon)
  • Timing Intelligence: "Peak hours, 9 AM - 5 PM" (coming soon)
  • Custom Signals: "Database migration required - Manual review needed" (coming soon)

Resource-Level Signals​

These signals analyse individual resources and appear in the Resources section, for example:

  • Routine Analysis: Shows change patterns for specific attributes for each resource. Allowing a user to see which attributes are changed frequently, and which are uncommon.
  • Change Type: "High risk", "First time changing" indicators for individual resources (coming soon)
  • Policy Compliance: "Security group allows 0.0.0.0/0 on port 22" (coming soon)

Currently, Routine Analysis is fully available at both levels, with other signal types in development.

How Signals Works​

Signals uses a scoring system from -5 to +5, where:

  • -5: High risk signals that indicate potential deployment concerns
  • 0: Neutral baseline
  • +5: Positive confidence signals that indicate safer deployments

Signal Score

Available Signals​

Routine Analysis​

Signals analyses the historical modification patterns of your infrastructure to determine how routine or exceptional your current changes are. Using advanced statistical analysis, it detects genuine behavioral patterns in your deployment history rather than just simple averages.

What it does:

  • Examines modification frequency over time for each resource
  • Automatically detects stable periods of consistent deployment behavior
  • Identifies when deployment patterns change significantly
  • Provides precise context like "2 events/day for the last 1 week and 1 day" or "15 events/day for the last 2 weeks and 3 days"

How it interprets patterns:

  • Frequent, consistent changes → Positive signals (well-established deployment patterns)
  • Infrequent or first-time changes → Negative signals (require additional attention)
  • Pattern breaks → Flags when current changes deviate from established routines

You can learn more about how routine analysis works in detail in the Routine Signal documentation.

Accessing Signals​

In Your Pull Request​

When using Overmind's GitHub Action integration, you'll see a Signals summary directly in your PR comments with a quick confidence assessment and "View Signals" link for detailed analysis.

Github PR comment with routine changes

Signals Page Interface​

Navigate to the Signals tab in your change to see the full analysis interface, which is organized into two main sections:

Top Section - Change-Level Overview​

  • Routine: "Recent infrastructure changes span multiple attributes, with modification frequencies ranging from first-time updates to daily deployments"

Bottom Section - Resources​

Individual resources are listed with confidence indicators and can be expanded for detailed analysis:

  • Click any resource (like api-server-nats or gateway) to see component-level breakdown
  • Each expanded resource shows routine analysis for components like metadata, spec, terraform_address
  • Historical patterns displayed as "2.0 events/day for the last 1 week and 1 day"

signals page

Interpreting Signals​

Visual Indicators​

The Signals interface uses clear visual cues:

  • Green indicators (🟢): Positive confidence - routine, established patterns
  • Red indicators (🔴): High attention needed - first-time changes or risk factors
  • Gray indicators (⚪): Neutral baseline signals

Reading Your Analysis​

High Confidence Deployments:

  • Most resources show green indicators with consistent patterns
  • Example: "Consistent routine deployments with 2.0 events/day over the past three weeks"

Attention Required:

  • Red indicators on resources or components
  • "First time changing this attribute" messages
  • "High risk" change type classifications

Component-Level Insights:

  • Expand any resource to see which specific parts are routine vs. exceptional
  • Different components may have different confidence levels
  • Focus review time on components showing red indicators or first-time changes

Best Practices​

For Development Teams​

Review Process:

  • Use Signals as input for deployment decisions, not absolute rules
  • Pay attention to first-time changes or unusual patterns
  • Consider both change-level and resource-level insights

For Platform Teams​

Setup:

  • Configure GitHub Action integration for automatic PR analysis
  • Set up appropriate midpoint thresholds for your deployment frequency
  • Train teams on interpreting Signals output

Monitoring:

  • Review Signals configuration periodically
  • Adjust sensitivity based on false positive/negative rates
  • Use historical analysis to refine thresholds

Interpreting Results​

When Signals analysis appears in your PR:

  1. Check Overall Confidence: Green indicates routine changes, red suggests careful review needed
  2. Review Risk Alerts: Pay attention to any highlighted exceptional changes
  3. Use Detail Link: Access full analysis for complex changes
  4. Make Informed Decisions: Combine Signals insights with your team's expertise

Next Steps​

Once you're familiar with Signals:

For additional help, visit our support documentation or join our Discord community.