A command-line interface designed for humans, agentic AIs and CI/CD to interact with the Dash0 observability platform.

The dash0 CLI is designed to be driven by AI coding agents as naturally as by humans. Its capabilities are discoverable via --help, alongside a comprehensive command reference with detailed flags, expected outputs, and ready-to-use workflow examples. Authentication and connection settings can be configured entirely through profiles and environment variables, avoiding the need to pass secrets as command-line arguments. Commands use consistent naming conventions and flags. Structured and parseable output formats (--output json, --output yaml, --output csv). Agent mode makes all of this automatic: JSON output, structured help, JSON errors, no prompts, and no colors — with zero configuration.

Tap the formula repository:

brew tap dash0hq/dash0-cli https://github.com/dash0hq/dash0-cli

Trust the tap. Homebrew 6.0 enforces tap trust by default for non-official taps; without this step, brew install and brew upgrade will refuse to load the formula.

brew trust dash0hq/dash0-cli

Install the CLI:

brew install dash0

Download pre-built binaries for your platform from the releases page. Archives are available for Linux, macOS and Windows across multiple architectures.

Use the dash0 CLI in your CI/CD workflows with the setup action:

steps:
  - uses: actions/checkout@v4

  - name: Setup Dash0 CLI
    uses: dash0hq/dash0-cli/.github/actions/setup@main  # You can use any git ref: @main, @v1.1.0, or @commit-sha
    # with:
    #   version: '1.1.0' # 1.1.0 is the earliest supported version

  - name: List dashboards
    env:
      DASH0_API_URL: ... # Find this at https://app.dash0.com/goto/settings/endpoints?endpoint_type=api_http
      DASH0_OTLP_URL: ... # Find this https://app.dash0.com/goto/settings/endpoints?endpoint_type=otlp_http
      DASH0_AUTH_TOKEN: ... # Get one from https://app.dash0.com/goto/settings/auth-tokens?auth_token_id=39d58aa9-b64e-464c-a675-cc3923085d6c ; keep the auth token in a GitHub secret!
      DASH0_DATASET: my-dataset # Leave empty for the `default` dataset
    run: dash0 dashboards list

The send-log-event action sends log events to Dash0 directly from your workflows. It is standalone: it installs the Dash0 CLI automatically if it is not already on PATH. If the setup action has already run in the same job, the existing installation is reused.

When used on its own, pass otlp-url and auth-token directly:

steps:
  - name: Send deployment event
    uses: dash0hq/dash0-cli/.github/actions/send-log-event@main
    with:
      otlp-url: ${{ vars.DASH0_OTLP_URL }}
      auth-token: ${{ secrets.DASH0_AUTH_TOKEN }}
      event-name: dash0.deployment
      body: 'Deployment completed'
      severity-number: '9'
      service-name: my-service
      deployment-environment-name: production
      deployment-status: succeeded

When the setup action has already created a profile, the connection parameters are inherited and do not need to be repeated:

steps:
  - name: Setup Dash0 CLI
    uses: dash0hq/dash0-cli/.github/actions/setup@main
    with:
      otlp-url: ${{ vars.DASH0_OTLP_URL }}
      auth-token: ${{ secrets.DASH0_AUTH_TOKEN }}

  - name: Send deployment event
    uses: dash0hq/dash0-cli/.github/actions/send-log-event@main
    with:
      event-name: dash0.deployment
      body: 'Deployment completed'
      severity-number: '9'
      service-name: my-service
      deployment-environment-name: production
      deployment-status: succeeded

docker run ghcr.io/dash0hq/cli:latest [command]

Multi-architecture images (linux/amd64, linux/arm64) are published to GitHub Container Registry.

Requires Go 1.22 or higher.

git clone https://github.com/dash0hq/dash0-cli.git
cd dash0-cli
make install

For the full command reference with detailed flags, output examples, and AI-agent workflows, see docs/commands.md.

The CLI resolves connection settings from profiles stored on disk, environment variables, and CLI flags, in that order. Profiles are the recommended way to manage credentials locally; environment variables are convenient for CI/CD and agentic workflows.

Configure API access using profiles. All profile fields are optional at creation time. Missing values can be supplied later via config profiles update or overridden at runtime with environment variables or CLI flags.

dash0 config profiles create dev \
    --api-url https://api.us-west-2.aws.dash0.com \
    --otlp-url https://ingress.us-west-2.aws.dash0.com \
    --auth-token auth_xxx

dash0 config profiles list
dash0 config profiles select dev
dash0 config show

You can find the API endpoint for your organization on the Endpoints page, under the API entry, and the OTLP HTTP endpoint under the OTLP via HTTP entry. Currently only HTTP OTLP endpoints are supported.

Profiles and the active-profile selection are stored on disk in ~/.dash0/:

The directory is created automatically when you create your first profile.

To store configuration elsewhere, set the DASH0_CONFIG_DIR environment variable:

export DASH0_CONFIG_DIR=~/.local/dash0
dash0 config profiles create dev --api-url https://api.us-west-2.aws.dash0.com

Agent mode optimizes every aspect of the CLI for machine consumption. Enable it explicitly with --agent-mode or DASH0_AGENT_MODE=true, or let it auto-activate when a known AI agent environment variable is detected:

When agent mode is active, the CLI:

• Defaults output to JSON — all data retrieval commands (list, get, query, config show, metrics instant) output JSON instead of tables, without needing -o json.
• Returns --help as structured JSON — flags, subcommands, aliases, and metadata are machine-parseable.
• Emits errors as JSON on stderr — {"error": "...", "hint": "..."} instead of colored text.
• Skips confirmation prompts — destructive operations (delete, remove) proceed without asking, equivalent to --force.
• Disables colored output — no ANSI escape codes in any output.

To explicitly disable agent mode (for example, when running inside an agent environment but wanting human-readable output), set DASH0_AGENT_MODE=0 or DASH0_AGENT_MODE=false. This overrides all other activation methods.

See the agent mode specification for the full priority order and details.

Apply asset definitions from a file, directory, or stdin. The input may contain multiple YAML documents separated by ---. Supported asset types: Dashboard, PersesDashboard, CheckRule, PrometheusRule (alerting and recording rules), SyntheticCheck, View, Dash0SpamFilter, and Dash0NotificationChannel.

dash0 apply -f assets.yaml
dash0 apply -f dashboards/
cat assets.yaml | dash0 apply -f -
dash0 apply -f assets.yaml --dry-run

Note: In Dash0, dashboards, views, synthetic checks and check rules are called "assets", rather than the more common "resources". The reason for this is that the word "resource" is overloaded in OpenTelemetry, where it describes "where telemetry comes from".

dash0 dashboards list
dash0 dashboards get <id>
dash0 dashboards get <id> -o yaml
dash0 dashboards create -f dashboard.yaml
dash0 dashboards create -f persesdashboard.yaml
dash0 dashboards update [id] -f dashboard.yaml
dash0 dashboards delete <id> [--force]

dash0 check-rules list
dash0 check-rules get <id>
dash0 check-rules get <id> -o yaml
dash0 check-rules create -f rule.yaml
dash0 check-rules create -f prometheus-rules.yaml
dash0 check-rules update [id] -f rule.yaml
dash0 check-rules delete <id> [--force]

Both apply and dashboards create also accept PersesDashboard CRD files. Both apply and check-rules create also accept PrometheusRule CRD files.

dash0 synthetic-checks list
dash0 synthetic-checks get <id>
dash0 synthetic-checks get <id> -o yaml
dash0 synthetic-checks create -f check.yaml
dash0 synthetic-checks update [id] -f check.yaml
dash0 synthetic-checks delete <id> [--force]

dash0 views list
dash0 views get <id>
dash0 views get <id> -o yaml
dash0 views create -f view.yaml
dash0 views update [id] -f view.yaml
dash0 views delete <id> [--force]

dash0 spam-filters list
dash0 spam-filters get <id>
dash0 spam-filters get <id> -o yaml
dash0 spam-filters create -f filter.yaml
dash0 spam-filters update [id] -f filter.yaml
dash0 spam-filters delete <id> [--force]

dash0 logs send "Application started" \
    --resource-attribute service.name=my-service \
    --log-attribute user.id=12345 \
    --severity-text INFO --severity-number 9

dash0 logs query
dash0 logs query --from now-1h --to now --limit 100
dash0 logs query --filter "service.name is my-service"
dash0 logs query --filter "otel.log.severity.range is_one_of ERROR WARN"
dash0 logs query --filter '[{"key":"service.name","operator":"is","value":"api"}]'
dash0 logs query -o csv
dash0 logs query --column time --column service.name --column body

The --filter flag also accepts JSON filter criteria copied from the Dash0 UI. See the filter syntax reference for the full list of operators.

dash0 spans send --name "GET /api/users" \
    --kind SERVER --status-code OK --duration 100ms \
    --resource-attribute service.name=my-service

dash0 spans query
dash0 spans query --from now-1h --to now --limit 100
dash0 spans query --filter "service.name is my-service"
dash0 spans query --filter "otel.span.status.code is ERROR"
dash0 spans query --filter '[{"key":"service.name","operator":"is","value":"api"}]'
dash0 spans query -o csv
dash0 spans query --column otel.span.start_time --column otel.span.duration --column "span name" --column http.request.method

See the filter syntax reference for the full list of operators.

dash0 traces get <trace-id>
dash0 traces get <trace-id> --from now-2h
dash0 traces get <trace-id> --follow-span-links
dash0 traces get <trace-id> -o json
dash0 traces get <trace-id> --column otel.span.start_time --column otel.span.duration --column "span name" --column otel.span.status.code

# Instant query — current request rate per service
dash0 metrics instant --promql 'sum by (service_name) (rate(http_server_request_duration_seconds_count[5m]))'

# Instant query with filters
dash0 metrics instant --filter 'service.name is my-service'

# Output as CSV with specific columns
dash0 metrics instant --promql 'sum by (service_name) (rate(http_server_request_duration_seconds_count[5m]))' -o csv --column value --column service_name

# List all teams
dash0 -X teams list

# Get team details (members + accessible assets)
dash0 -X teams get <id>

# Create a team
dash0 -X teams create "Backend Team" --color-from "#FF6B6B" --color-to "#4ECDC4"

# Add members to a team
dash0 -X teams add-members <team-id> <member-id-1> <member-id-2>

# List organization members
dash0 -X members list

# Invite a member (default role: basic_member)
dash0 -X members invite user@example.com

# Delete a member
dash0 -X members delete <member-id> --force

The api command calls any Dash0 API endpoint directly, reusing the active profile's connection settings. It is useful for endpoints that do not yet have a dedicated subcommand.

# GET — dataset auto-injected from the active profile
dash0 -X api /api/signal-to-metrics/configs

# POST with a payload from a file
dash0 -X api POST /api/signal-to-metrics/configs -f config.json

# Skip dataset injection for organization-level endpoints
dash0 -X api /api/organization/settings --dataset ""

The otlp proxy command runs a long-lived local OTLP forwarder. It accepts OTLP/HTTP on 127.0.0.1:4318 and OTLP/gRPC on 127.0.0.1:4317 and forwards every batch to Dash0 using the active profile's credentials. An OpenTelemetry SDK at default endpoint configuration connects without any environment variable change, collapsing the local-dev setup from a YAML-heavy Collector config to a single command.

# Just run it. SDK defaults already point at the proxy.
dash0 -X otlp proxy

# Print every forwarded record on stdout in collector-debug-exporter style.
dash0 -X otlp proxy --tail

# Pick non-default ports (e.g., when another local Collector holds the defaults).
dash0 -X otlp proxy --http-port 8318 --grpc-port 8317

# Tag every forwarded batch at the resource level so it is filterable in Dash0.
dash0 -X otlp proxy \
    --resource-attribute developer=alice \
    --resource-attribute deployment.environment.name=local

The proxy exits on Ctrl-C (or SIGTERM) after draining in-flight work within a 5-second deadline. On startup, if either default port is already in use, the proxy exits non-zero with an actionable error that names the holding process. See docs/commands.md for the full reference, including the decoration flags (--scope-attribute, --log-attribute, --span-attribute, --metric-attribute, --scope-name, --scope-version), the agent-mode event schema, and the failure-mode classification.

The list and get commands for assets support multiple output formats via -o:

• table (default): Compact view with essential columns (name and ID)
• wide: Similar to table, with additional columns (dataset, origin, and URL)
• json: Full asset data in JSON format
• yaml: Full asset data in YAML format
• csv: Comma-separated values with the same columns as wide, suitable for piping and automation

The update and apply commands show a unified diff of changes.

The logs query, spans query, and traces get commands support a different set of formats via -o:

• table (default): Columnar output (logs query shows timestamp, severity, and body; spans query shows timestamp, duration, name, status, service, and trace ID; traces get shows a hierarchical span tree)
• json: Full OTLP/JSON payload
• csv: Comma-separated values

In agent mode, all data retrieval commands default to JSON without needing -o json.

Enable tab completion for your shell:

Bash (requires bash-completion):

source <(dash0 completion bash)
# Permanent (macOS): dash0 completion bash > $(brew --prefix)/etc/bash_completion.d/dash0

Zsh:

source <(dash0 completion zsh)
# Permanent (macOS): dash0 completion zsh > $(brew --prefix)/share/zsh/site-functions/_dash0

Fish:

dash0 completion fish | source
# Permanent: dash0 completion fish > ~/.config/fish/completions/dash0.fish

See CONTRIBUTING.md for development instructions.