Skip to content

cli-guide.md

Latest commit

 

History

History
130 lines (92 loc) · 4.21 KB

cli-guide.md

File metadata and controls

130 lines (92 loc) · 4.21 KB
Raw

CLI guide

The devops-ai command is a thin Typer wrapper over the shared AnalysisEngine. All real work happens in the engine; the CLI only handles input/output and rendering.

devops-ai --help

Commands at a glance

Command Purpose
devops-ai analyze <file> Analyze a log, manifest, Terraform file, or output
devops-ai explain <error> Explain a known error from the knowledge base
devops-ai validate <file> Validate a YAML / Kubernetes / Terraform document
devops-ai list List the error signatures the toolkit knows about
devops-ai serve Run the FastAPI REST API (needs the api extra)
devops-ai version Print the installed version

analyze

Analyze a file, or pipe content via stdin using -.

devops-ai analyze nova.log
kubectl get pods | devops-ai analyze -
cat terraform-apply.txt | devops-ai analyze -

Options:

Option Description
--tech, -t Technology hint, e.g. kubernetes, terraform, openstack
--enrich Add an LLM narrative (requires a configured provider)
--provider Override the provider: anthropic | openai | gemini | ollama
--json Emit machine-readable JSON instead of the Rich rendering

Examples:

devops-ai analyze app.log --tech kubernetes
devops-ai analyze app.log --json | jq '.root_causes[0]'
devops-ai analyze app.log --enrich --provider anthropic

Exit code: 0 if at least one signature/root cause matched, 1 if nothing matched, 2 on input errors (missing file, empty stdin). This makes it easy to gate CI steps on a match.

explain

Explain a named error without any input file.

devops-ai explain CrashLoopBackOff
devops-ai explain "exit code 137"
devops-ai explain ImagePullBackOff --json

Returns the title, summary, root causes, diagnostic commands, fixes, references, and best practices for the best-matching signature. Exit code 0 when matched, 1 otherwise.

validate

Read-only validation of a manifest or Terraform file. It never applies anything.

devops-ai validate deploy.yaml
cat deploy.yaml | devops-ai validate -
devops-ai validate main.tf --json

Exit code 0 when valid, 1 when validation issues are present. See Output format for the issue structure.

list

Show the signatures in the knowledge base, optionally filtered by technology.

devops-ai list
devops-ai list --tech kubernetes
devops-ai list -t terraform

serve

Start the REST API (requires pip install 'devops-ai-toolkit[api]').

devops-ai serve                       # 127.0.0.1:8000 by default
devops-ai serve --host 0.0.0.0 --port 8080

See the REST API guide.

version

devops-ai version

Piping and composition

Because analyze and validate accept stdin (-) and support --json, the CLI composes well with standard tooling:

# Triage the most recent crashed pod's previous logs
kubectl logs my-pod --previous | devops-ai analyze - --tech kubernetes

# Fail a CI job if a Terraform plan trips a known error
terraform plan 2>&1 | devops-ai analyze - --tech terraform || echo "known issue detected"

# Extract just the top suggested fix as JSON
devops-ai analyze app.log --json | jq -r '.suggested_fixes[0].title'

Configuration

The CLI honours the same environment variables as the rest of the toolkit (provider selection, API keys, timeouts, input size). See Configuration.

See also