Features

Everything you need to understand, optimize, and report the energy cost of your infrastructure.

Multi-Format Intelligence

Native LSP support for every infrastructure format. Hover for energy costs, jump to dependencies, inline blast radius.

HCL / Terraform

Full semantic understanding of resources, data sources, modules, and state. Energy cost in plan output.

resource "load_balancer" "main" {
  # LSP: 3 services depend on this resource
  # Blast radius: 7 endpoints affected
  # Energy: 0.8 J/hr steady-state
  type     = "application"
  internal = false
}

Supported Formats

Native parsing and validation for the entire infrastructure ecosystem.

✓Terraform HCL — plan, apply, state
✓Kubernetes YAML — all resource types
✓Docker / Compose — multi-stage builds
✓Helm — templates and values
✓Ansible — playbooks and roles
✓Nix — derivations and flakes
✓GitHub Actions / GitLab CI — workflows
✓Shell scripts — static analysis

Infrastructure Graph

Your entire infrastructure stored as a queryable graph in JouleDB. Dependencies, blast radius, drift — all traversable.

$ devops graph query "path from ingress to database"

ingress-nginx
  -> api-gateway     0.12 J/hr
  -> auth-service    0.34 J/hr
  -> user-api        0.28 J/hr
  -> postgres-primary 1.47 J/hr

Total path energy: 2.21 J/hr
Critical path: yes (single point of failure at auth-service)

Dependency Visualization

See every upstream and downstream dependency. Filter by namespace, team, or energy cost.

Blast Radius Analysis

Before changing a resource, see exactly what breaks. Includes transitive dependencies.

Cross-Format Links

Terraform resource linked to K8s deployment linked to Docker image. One unified graph.

Pipeline Energy Metering

Every CI/CD stage reports energy consumed. Track regressions. Set energy budgets. Ship energy receipts.

$ devops pipeline energy --last 5

Run #142  build 1,247 mJ | test 892 mJ | deploy 156 mJ  total 2,295 mJ
Run #141  build 1,312 mJ | test 901 mJ | deploy 148 mJ  total 2,361 mJ
Run #140  build 3,847 mJ | test 894 mJ | deploy 152 mJ  total 4,893 mJ !! regression
Run #139  build 1,198 mJ | test 887 mJ | deploy 161 mJ  total 2,246 mJ
Run #138  build 1,205 mJ | test 890 mJ | deploy 155 mJ  total 2,250 mJ

Container Analysis

Layer-by-layer energy breakdown for every container image. Identify the expensive layers and optimize.

$ devops container analyze api:latest

Layer 1  FROM debian:bookworm-slim   12.4 mJ  (base)
Layer 2  apt-get install libssl3     84.2 mJ  (deps)
Layer 3  COPY target/release/api     2.1 mJ   (binary)
Layer 4  RUN chmod +x /app           0.3 mJ   (meta)

Total build energy: 99.0 mJ
Suggestion: Layer 2 is 85% of cost. Consider static linking.

Cloud Energy Translation

Translate $/hr cloud pricing into joules/operation. Compare instance types by energy efficiency, not just cost.

$ devops cloud compare --workload api-server

Instance Type        $/hr    J/req    Efficiency
c7g.medium (Arm)     $0.034  0.34 J   best
c6i.large  (x86)     $0.085  0.71 J   2.1x worse
m5.large   (GP)      $0.096  1.00 J   2.9x worse

Switching to c7g.medium saves 66% energy and 60% cost.

Secret Detection

Scan infrastructure files for hardcoded secrets, API keys, and credentials. Pre-commit hooks block secrets before they reach the repository.

$ devops scan --secrets .

FOUND 3 potential secrets in 2 files:

  deploy/values.yaml:14    API_KEY = "sk-..."  (high confidence)
  terraform/main.tf:42     password = "..."    (high confidence)
  .env.example:7           TOKEN = "..."       (medium confidence)

Blocked: commit rejected. Use vault references instead.

GitOps Native

Declarative infrastructure with continuous reconciliation. Every sync reports energy status.

$ devops gitops status

Cluster: production
  synced    api-deployment          (3 replicas, 14.2 J/hr)
  synced    redis-statefulset       (1 replica, 8.7 J/hr)
  drifted   nginx-configmap         (manual edit detected)
  synced    postgres-secret         (vault-backed)

Overall: 3/4 synced | Total cluster energy: 37.1 J/hr

16-Level Cascade

Infrastructure queries resolved at the cheapest possible level. Cache and index handle most queries in under 1 mJ. LLM is the escape hatch, not the default.

Level     Engine            Energy     Use Case
L0        Cache             0.001 mJ   Recent queries
L1        Tantivy Index     0.1 mJ     Full-text search
L1.25     Graph RAG         0.5 mJ     Relationship traversal
L2        Federation        2 mJ       Cross-source search
L2.5      Neural Rerank     5 mJ       Relevance scoring
L3        LLM (cheap)       50 mJ      Simple reasoning
L4        LLM (expensive)   500 mJ     Complex analysis
...
L10       Deep Verification 5,000 mJ   Formal proof

Drift Detection

Continuous comparison of declared state vs. live state. Every drift event includes the energy cost delta and root cause.

$ devops drift check --continuous

Watching 47 resources across 3 clusters...

[12:04:22] DRIFT detected: nginx-ingress
  declared: replicas=2, actual: replicas=3
  energy delta: +4.7 J/hr
  cause: manual kubectl scale (user: ops-bot)

[12:04:22] DRIFT detected: redis-config
  declared: maxmemory=256mb, actual: maxmemory=512mb
  energy delta: +2.1 J/hr
  cause: runtime reconfiguration

Compliance Dashboard

Generate compliance reports from real energy measurements. ISO/IEC 21031 SCI, EU CSRD Scope 3, GHG Protocol, and PUE — all from actual data.

$ devops compliance report --format sci

Software Carbon Intensity Report
Period: 2026-03-01 to 2026-03-20

Functional unit: API request
Energy (E):           0.47 J/request (measured)
Carbon intensity (I): 142 gCO2/kWh (grid-average, us-east-1)
Embodied (M):         0.003 gCO2/request (amortized)

SCI = 0.019 gCO2/request

Exported: sci-report-2026-03.json