InteliG
Sign Up

Code Intelligence

Everything that happens in your repositories, analyzed and surfaced as engineering intelligence.

What Code Intelligence Is

Code Intelligence is InteliG's pillar for understanding everything in your repositories. Commits, pull requests, deployments, KPIs, contributor behavior, codebase health -- all ingested, analyzed, and turned into actionable intelligence.

This is not activity tracking. Cognis processes raw git data through AI scoring pipelines and engineering profiles to produce quality scores, risk assessments, delivery metrics, and organizational insights. The goal is a complete, honest picture of how your engineering organization builds software.

Performance (Insights)

The Insights page computes an Engineering Intelligence Score from 0 to 100, built from four pillar scores:

Throughput

How much code is being integrated?

Cycle Time

How fast are changes being merged?

Quality Risk

Where are the risks?

Engagement

Who is building?

Each pillar has a score, a status (excellent, strong, good, watch, concern, critical), a trend direction (up, down, stable), and a trend percentage. Optional insights include severity, summary, and a recommendation.

The page also generates an Intelligence Story -- an AI-written narrative of your engineering state with a top recommendation, including the model and timestamp metadata for transparency.

KPIs

KPIs are available at three scopes via toggle tabs: Organization, Team, and Contributor. All scopes share a selectable date range.

Core Metrics (all scopes)

  • → Commits -- total commit count
  • → Pull Requests -- total PR count
  • → Merge Velocity -- average merge speed
  • → PR Cycle Time -- average time from open to merge
  • → Review Responsiveness -- how quickly reviews happen
  • → Review Coverage -- percentage of PRs that received reviews

Quality Metrics

  • → Rework Rate -- percentage of PRs requiring changes before merge
  • → First-Pass Approval Rate -- percentage of PRs approved on first review
  • → Delivery Consistency -- standard deviation of cycle times (lower is more predictable)
  • → Review Reciprocity -- ratio of reviews given to reviews received (1.0 is balanced)

Phase 2 Metrics

  • → Work Type Distribution -- breakdown of commit classifications (Feature, Bugfix, Refactoring, etc.)
  • → Repository Ownership -- top repos with ownership percentage per contributor
  • → Review SLA Compliance -- percentage of reviews completed within 24 hours
  • → Weekly Velocity -- average commits per week
  • → Context Switching Index -- distinct repos worked on per contributor

Organization scope adds deploy count, idle time, merge rate, output change percentage, and an overall rating (Excellent, Moderate, At Risk). Contributor scope adds a performance rating and performance notes.

Repositories

The Repositories page shows all connected repositories with their status and provider type. You can confirm or deconfirm repositories for tracking, and trigger a resync with progress feedback. This is where you manage which repositories feed into your intelligence pipeline.

Delivery

Tracks deployments using data from the GitHub Deployments API. Each deployment records its environment, ref, SHA, creator, and full status lifecycle: Pending, Queued, In Progress, Success, Failure, Error, Inactive.

  • → Repository view -- last production deploy, success/failure counts, success rate, distinct environments, contributors shipping, cadence classification (daily, weekly, sporadic, inactive), skips-staging detection
  • → Contributor view -- production and non-production deploy counts, distinct repos, success rate
  • → Environment view -- per-environment counts, repos using and bypassing each environment, average hours before promotion, promotion flow tracking (staging to production with promoted SHA counts)

Deployment frequency is bucketed into time series with total, production, non-production, success, and failure counts per period. DORA-aligned metrics are derived from this data.

Engineering Org

A composite page with two tabs for managing the people side of your engineering organization.

Contributors

Manage git contributors: assign roles and positions, merge duplicate identities (automatic detection plus manual merge), manage identity aliases (multiple git identities mapped to one person), toggle visibility in analytics, and track idle status.

Teams

Create and manage teams. Team-level KPIs inherit from the KPI system, giving you the same metric depth at the team scope.

Codebase Health

Portfolio-level repository intelligence across four lenses. Each repository gets an overall status (healthy, moderate, at risk, dormant) based on the worst score across all lenses.

Health

Repository activity and freshness. Status: active, moderate, or dormant.

Hygiene

Stale branch count and branch debt. Status: clean, watch, or dirty.

Flow

Open PR count and oldest PR age. Status: flowing, slow, or blocked.

Ownership

Top contributor share and active contributor count. Status: distributed, concentrated, or bus factor risk.

Deep Dives

Explore the AI analysis layers that power Code Intelligence:

Commit Intelligence

AI analysis of every commit -- engineering quality scoring, risk assessment, classification, effort estimation, and AI attribution detection.

PR Intelligence

Full PR lifecycle tracking -- state management, review tracking, cycle time computation, and the metrics that feed into your KPIs.

Integrations

Code Intelligence ingests data from your git provider. GitHub is the primary integration with full webhook support, HMAC signature verification, OAuth, and GitHub App installation. Additional providers are defined and supported:

  • → GitHub -- primary provider, full webhook and real-time sync
  • → GitLab
  • → Bitbucket
  • → Azure DevOps
  • → AWS CodeCommit

Data sync happens through three channels: initial historical import (configurable from 30 days to 12 months), real-time webhooks, and nightly reconciliation jobs. Webhook processing includes deduplication by delivery ID, async Kafka processing, retry scheduling, and dead letter topic handling.