CLI

Cloud

• everr cloud login — authenticate with your Everr account
• everr cloud logout — sign out
• everr cloud query "<SQL>" — run a read-only SQL query against your cloud telemetry and CI data
  • --format <json|ndjson|table> output format (default: table in a terminal, ndjson otherwise)

Setup

• everr setup — run the guided setup wizard, including optional skills installation
• everr init — import runs for the current repository

Skills

• everr skills list [--project|--global] [--agent <codex|claude-code|cursor|all>] — list bundled Everr skills
• everr skills install [--all] [--project|--global] [--agent ...] [--force] [--dry-run] — install all bundled skills; omitting --all in a terminal asks for confirmation
• everr skills update [SKILL]... [--project|--global] [--agent ...] [--dry-run] — update installed bundled skills
• everr skills uninstall [SKILL]... [--all] [--project|--global] [--agent ...] [--yes] [--dry-run] — uninstall bundled skills

CI/CD status

• everr ci status — show the current commit's pipeline state (state, active, completed)

  • --commit <sha> target a specific commit
  • --run-id <id> target a specific run
  • --branch <name> / --repo <repo> override repository context

• everr ci watch — wait until the current HEAD commit finishes; exits non-zero if any run fails

  • --commit <sha> target a specific commit
  • --attempt <n> target a specific retry attempt
  • --run-id <id> target a specific run
  • --fail-fast exit immediately when any run fails

• everr ci runs — list recent runs

  • --branch <name> / --current-branch filter by branch
  • --conclusion <success|failure|cancellation> filter by outcome
  • --workflow-name <name> filter by workflow
  • --run-id <id> fetch a specific run
  • --from <expr> / --to <expr> restrict to a time range (accepts date math, e.g. now-7d)

• everr ci show <trace_id> — show jobs and steps for a run

  • --failed show only failed jobs and their failed steps

• everr ci logs <trace_id> --job-name <job> --step-number <n> — print step logs (last 1000 lines by default)

  • --job-name <job> / --job-id <id> select a job (required)
  • --step-number <n> / --log-failed select a step (exactly one required)
  • --tail <n> how many lines from the bottom
  • --limit <n> / --offset <n> raw paging
  • --egrep <pattern> keep only lines matching a re2 regex (exits non-zero if none match)
  • --color preserve ANSI color codes (stripped by default)

Collection-style commands accept --limit <n> and --offset <n> for pagination.

Local telemetry

• everr local start — start the local OpenTelemetry collector
  • --quiet suppress the collector URL output after startup
• everr local status — check whether the local OpenTelemetry collector is running and print local OTLP/SQL URLs
• everr local query "<SQL>" — query local telemetry with read-only SQL
  • --format <json|ndjson|table> output format (default: table in a terminal, ndjson otherwise)
• everr wrap -- <command> — run a command normally and mirror stdout/stderr into local telemetry
  • Requires a running collector; if unavailable, the command is not run
  • Preserves the wrapped command's non-zero exit code

Dashboards (gitops)

• everr apply <dir> — reconcile a directory of resource definitions (.yaml/.yml/.json) into Everr. By convention the directory is everr/ at your repo root, holding flat, kind-suffixed files: *.dashboard.yaml, *.alert.yaml, *.runbook.yaml (the legacy *.notebook.yaml suffix is still recognized). Apply discovers kind: Dashboard, kind: AlertRule, and kind: Runbook documents under the directory, routing by the kind: field (the suffix is a naming convention only). The directory must contain an everr.yaml (or everr.yml) at its root declaring a stable repoid (repoid: "2f8e3f90-..."); apply errors without it. The repoid is the authoritative reconcile scope — the apply ownership boundary — and is the only key the manifest accepts. Within that repoid it makes the live resources match the tree: new files are created, changed files updated, and removed files deleted. Dashboards may set an optional metadata.project (defaults to default) to namespace identity and URL. See Observability as code.

  • <dir> directory to read resource files from (recursively), e.g. ./everr; must contain an everr.yaml manifest at its root; any subdirectories become folder paths
  • --preview [NAME] apply into a preview namespace instead of the live state and print a shareable link; NAME defaults to the current git branch (pass it explicitly in CI or on a detached HEAD). Previews apply without confirmation, alert rules in them evaluate but never notify, and each preview expires automatically 7 days after its last apply
  • --dry-run compute and print the diff without writing
  • --yes, -y skip the confirmation prompt (required in non-interactive contexts such as CI or pipes)

  Authentication: apply uses your everr cloud login session by default, and prints the destination organization. In GitHub Actions, install the CLI with everr-action before calling everr apply. For CI or non-interactive use, set EVERR_API_KEY to an organization API key with the apply scope. With changes to apply, it shows a confirmation prompt before writing; pass --yes (or run with EVERR_API_KEY in a non-interactive context) to skip it.

  • EVERR_API_KEY an organization API key (prefix ek_) with the apply scope, used as a bearer token; when set it takes precedence over the session. The legacy EVERR_API_TOKEN is accepted as a deprecated alias.
  • EVERR_API_URL API base URL (falls back to the URL persisted by a prior everr cloud login)

Other

• everr uninstall — remove all config files, log out, and print the command to remove the CLI binary