Data Sources & Adapters

The more sources connected, the richer the pattern detection. Every tool your team already uses can feed into Evolution Engine.

How It Works

Evolution Engine monitors your development process by analyzing signals from multiple sources and detecting cross-family patterns that no single tool can see.

Your existing tools          Evolution Engine              What you get
─────────────────           ─────────────────             ──────────────
Git history          ──➔    ┌─────────────────┐
CI/CD (Actions)      ──➔    │  Signal Engine   │──➔  Cross-family patterns
Dependencies (npm)   ──➔    │  (Phase 2)       │──➔  Risk advisories
Security (Snyk)      ──➔    │                  │──➔  Drift detection
Deployments          ──➔    │  Pattern Engine  │──➔  Trend reports
Monitoring           ──➔    │  (Phase 4)       │──➔  Investigation prompts
Quality gates        ──➔    └─────────────────┘

Integration Levels

Evolution Engine delivers value at every level. Each additional data source unlocks new pattern families.

Level 1 — Git Only Free

What you need: A git repository. That's it.

What you get:

Commit-level signals: files touched, file dispersion, change locality, co-change novelty
Baseline comparison: “This commit touched 12 files — about 3x more than usual”
Risk advisories for unusually large, scattered, or novel changes

evo analyze .

Level 2 — Git + Dependencies Free

What you need: A supported lockfile (package-lock.json, go.sum, Cargo.lock, Gemfile.lock, etc.).

What you get: Everything in Level 1, plus:

Dependency count and depth tracking
Cross-family pattern: “Commits that change dependencies tend to touch 40% more spread-out files”
Supply chain drift detection

Level 3 — Git + CI Pipeline Pro

What you need: GitHub Actions, GitLab CI, or CircleCI.

Setup:

# GitHub Actions
export GITHUB_TOKEN=$(gh auth token)

# GitLab CI
export GITLAB_TOKEN=glpat-...

What you get: Everything in Level 2, plus:

Build duration and failure rate tracking
Cross-family patterns: CI-triggering commits vs. code dispersion
Correlation between change size and build time trends

Level 4 — Git + CI + Deployments Pro

What you need: GitHub Releases or GitLab Releases.

What you get: Everything in Level 3, plus:

Release cadence, pre-release tracking, asset count monitoring
Cross-family patterns: release commits vs. file dispersion
Deployment velocity drift detection

Level 5 — Full Stack Pro + Adapters

What you need: Any combination of the adapters below.

Adapter	Signal Family	Example Cross-Pattern
Sentry	Error Tracking	“Dependency updates correlate with error spikes”
Snyk / Semgrep	Security	“High-dispersion commits introduce more vulnerabilities”
SonarQube	Quality Gate	“Low locality commits have more code smells”
Datadog	Monitoring	“High-dispersion deploys correlate with latency spikes”
Jira / Linear	Work Items	“Rework items have higher co-change novelty”

The Value Equation

Each new source doesn't add linearly — it multiplies the pattern space.

Connected Sources	Families	Cross-Family Combinations
Git only	1	0 (single-family trends only)
Git + deps	2	1
Git + deps + CI	3	3
Git + deps + CI + deploy	4	6
Full stack (6+)	6+	15+

Setting Up Tokens

Token-based adapters (CI, deployments, security) need an API token to fetch data. Set these as environment variables before running evo analyze.

GitHub

# Use the GitHub CLI to get your token (easiest)
export GITHUB_TOKEN=$(gh auth token)

# Or create a personal access token at github.com/settings/tokens
# Required scopes: repo (read), actions (read)
export GITHUB_TOKEN=ghp_...

GitLab

# Create a personal access token at gitlab.com/-/user_settings/personal_access_tokens
# Required scopes: read_api
export GITLAB_TOKEN=glpat-...

Sentry

# Create an auth token at sentry.io/settings/auth-tokens/
export SENTRY_AUTH_TOKEN=sntrys_...
export SENTRY_ORG=your-org
export SENTRY_PROJECT=your-project

Verify your setup

# See what's connected and what could be added
evo sources

Troubleshooting: Token Set but No Data?

If your report shows "Token set but no data collected" or "Connected — No Data", here's what to check for each adapter.

CI / Build (GitHub Actions, GitLab CI, CircleCI)

Need pipeline runs. EE reads the last 90 days of CI runs. If the repo has no workflow runs in that window, there's nothing to analyze.
Check: Go to your repo's Actions tab (GitHub) or CI/CD → Pipelines (GitLab) and confirm runs exist.
Token scopes: GitHub needs actions:read. GitLab needs read_api.
Requires Pro for API-based CI adapters. See pricing.

Deployments (GitHub Releases, GitLab Releases)

Need published releases. Git tags alone are not enough — you need actual Releases with release notes.
Check: Go to your repo's Releases page and confirm published releases exist.
Requires Pro.

Security (Dependabot)

Need Dependabot enabled. Go to repo Settings → Code security and analysis → enable Dependabot alerts.
No alerts = no data. If Dependabot is enabled but finds no vulnerabilities, there's nothing to report (which is good!).
Requires Pro.

Error Tracking (Sentry)

Need events in the project. EE queries the last 90 days of Sentry issues.
Check env vars: All three are required: SENTRY_AUTH_TOKEN, SENTRY_ORG, SENTRY_PROJECT.
Requires Pro.

Testing & Coverage (JUnit XML, Cobertura XML)

File-based — no token needed. These adapters look for report files in your repo.
Common paths: **/TEST-*.xml, **/junit.xml, **/coverage.xml, **/cobertura.xml.
CI artifacts: If your test/coverage reports are generated during CI but not committed, they won't be found locally. Either commit them or run evo analyze in CI where the files exist.

Platform Mismatch

If you see "Platform Mismatch", it means your token doesn't match the git remote. For example, GITLAB_TOKEN is set but the repo's remote points to github.com. Set the correct token for your hosting platform, or unset the mismatched one.

Quick diagnostic: Run evo sources to see exactly which adapters are connected, which need tokens, and which need a Pro license.

Check What More Sources Would Add

Entirely optional. Evolution Engine works standalone with just git. But if you're curious:

evo sources --what-if datadog sentry

You can compare reports side by side — run once without adapters, once with — and see if the additional patterns are worth it for your team.

FAQ

Do I need to replace my existing tools?

No. Evolution Engine sits alongside your tools, not instead of them. It ingests their output as signals and finds patterns across all of them.

What if I only have git?

That's enough to start. Git-only analysis detects commit-level anomalies. Many teams start here and add sources over time.

Does my code leave my machine?

No. Evolution Engine runs locally. Code, events, and signals stay on your machine. The only optional network calls are for API-based adapters and the opt-in community knowledge base.

How is this different from Snyk / SonarQube / Datadog?

Those tools analyze individual events at a point in time. Evolution Engine analyzes the process over time — detecting gradual drift, cross-family correlations, and patterns that only emerge across hundreds of commits. Their output can feed into EE as additional signal families.