Install the agent
We recommend installing the nullplatform agent via Helm. It’s the best fit for production clusters and CI/CD automation.
If you’re just testing locally, you can also run the agent directly with the np-agent binary.
Choose your installation method
- Install via Helm. (Recommended for production and automation)
- Use the CLI script to install in one step for serverless instances or testing.
Prerequisites
Before you start, make sure you have:
- A Kubernetes cluster
- Helm 3+ (if you’re installing with Helm)
- An API key with the agent and ops roles. See Authenticate the agent
Install with Helm
The following instructions rely on the nullplatform Helm charts repo.
1. Add the Helm repository
helm repo add nullplatform https://nullplatform.github.io/helm-charts
helm repo update
2. Set your agent values and install
The minimum values you’ll need to install is an API key, the agent repo, and tags.
export NP_API_KEY=<your_api_key>
export AGENT_TAGS=<key1:value1,key2:value2>
export AGENT_REPO=<https://git-provider/your-org/your-repo.git#your_branch>
💡 Tip: Check the logs to confirm that your
AGENT_REPOwas cloned correctly. See the Variables reference section below for more information and examples.
Then run helm install command with those values:
helm install nullplatform-agent nullplatform/nullplatform-agent \
--set configuration.values.NP_API_KEY=$NP_API_KEY \
--set configuration.values.AGENT_REPO=$AGENT_REPO \
--set configuration.values.TAGS="$AGENT_TAGS"
Quick install with the np-agent script
Use this for local testing or troubleshooting. For Kubernetes and production installs, use Helm.
If you just want to get the agent up and running without configuring Helm, you can use the quick-install script.
This method downloads the np-agent binary and lets you run it directly from your terminal.
1. Download and install the agent
This downloads the latest np-agent binary to your local system (usually to ~/.local/bin/np-agent).
curl https://cli.nullplatform.com/agent/install.sh | bash
Once complete, you’ll be able to call np-agent from your terminal.
2. Run the agent
Start the agent by passing in your API key, tags, and any additional parameters:
np-agent \
--api-key=$NP_API_KEY \
--runtime=host \
--tags=key:value \
--command-executor-env=NP_API_KEY="\"$NP_API_KEY\"" \
--command-executor-debug \
--webserver-enabled
Note: np-agent also reads NP_API_KEY and NP_LOG_LEVEL from your environment if they’re set.
Here’s a quick rundown of those flags:
--api-key: Authenticates the agent with nullplatform (your API key from the prerequisites).--runtime: Specifies where the agent runs (hostmeans directly on your system).--tags: Adds metadata tags to the agent. Format them as comma-separatedkey:valuepairs.--command-executor-env: Passes environment variables to executed commands.--command-executor-debug: Enables debug output for command execution.--webserver-enabled: Starts the agent’s built-in webserver.
np-agent --help
Optional flags (recommended)
-
Log verbosity: Choose how much detail the agent prints. Helpful for troubleshooting.
--log-level=INFOYou can choose from:
ERROR– Only errorsWARNING– Warnings and errorsINFO– General run info (default)DEBUG– Extra-detailed debug info
-
Pretty-print logs: Makes console output easier to read: colorized and nicely formatted JSON.
--log-pretty-print
Verify installation
Let’s make sure everything is working properly.
Check the agent in the UI
- In the platform UI, go to Platform settings > Agents.
- You can filter by tags or agent ID to find your new agent.
You should see your new agent listed by that ID.
Variables (reference & examples)
This section is a quick lookup for the values you pass to the agent chart. For more details, see the nullplatform Helm charts repo.
np-agent directly?You’ll pass the same data as flags instead of Helm values. np-agent also reads NP_API_KEY and NP_LOG_LEVEL from your environment if they’re set.
NP_API_KEY→--api-keyAGENT_REPO→--command-executor-git-command-reposTAGS→--tagsNP_LOG_LEVEL→--log-level
Required values
These are the minimum values you need to install the chart.
| Value | What it does | Helm path | Example |
|---|---|---|---|
| API key | Authenticates the agent with nullplatform | configuration.values.NP_API_KEY | NP_API_KEY="npk_..." |
| Agent repo | Tells the agent where to read scopes and actions | configuration.values.AGENT_REPO | AGENT_REPO="https://git-provider/org/repo.git#main" |
| Tags | Adds metadata for filtering in the UI | configuration.values.TAGS | TAGS="environment:dev,team:platform" |
Optional values
Use these when you want to organize or tune the agent.
| Value | What it does | Helm path | Example |
|---|---|---|---|
| Log level | Changes verbosity | configuration.values.NP_LOG_LEVEL | NP_LOG_LEVEL="INFO" |
AGENT_REPO formats
The agent repo can be a single repo or a comma-separated list. Use the #branch suffix to select a branch.
Single public repository
AGENT_REPO="https://git-provider/your-org/your-repo.git#your_branch"
Single private repository (with token)
AGENT_REPO="https://<git-token>@git-provider.com/your-org/private-repo.git#your_branch"
Multiple repositories (comma-separated, no spaces)
AGENT_REPO="https://git-provider/your-org/your-repo.git#your_branch,https://<git-token>@git-provider.com/your-org/private-repo.git#your_branch"
For the full list of values and defaults, check the nullplatform Helm charts repo.
Next steps
Installing the agent only registers it. To trigger actions, create an agent notification channel after install. See Notification channels.