kubescape/httphandler

Kubescape HTTP Handler Package

Running kubescape will start up a web-server on port 8080 which will serve the following API's:

Trigger scan

• POST /v1/scan - triggers a Kubescape scan. The server will return an ID and will execute the scanning asynchronously. The request body should look as follows.
• • wait=true: scan synchronously (return results and not ID). Use only in small clusters or with an increased timeout. Default is wait=false
• • keep=true: do not delete results from local storage after returning. Default is keep=false

Response:

{
  "id": <str>,                      // scan ID
  "type": "busy",                   // response object type
  "response": <message:string>      // message indicating scanning is still in progress
}

When scanning was triggered with the wait=true query param, the response is like the /v1/results API response

Get results

• GET /v1/results - request kubescape scan results
• • query id=<string> -> request results of a specific scan ID. If empty will return the latest results
• • query keep=true -> keep the results in the local storage after returning. default is keep=false - the results will be deleted from local storage after they are returned

Response:

When scanning was done successfully

{
  "id": <str>,                      // scan ID
  "type": "v1results",              // response object type
  "response": <object:v1results>    // v1 results payload
}

When scanning failed

{
  "id": <str>,                  // scan ID
  "type": "error",              // response object type
  "response": <error:string>    // error string
}

When scanning is in progress

{
  "id": <str>,                    // scan ID
  "type": "busy",                 // response object type
  "response": <message:string>    // message indicating scanning is still in progress
}

Check scanning progress status

Check the scanning status - is the scanning in progress or done. This is meant for a waiting mechanize since the API does not return the entire results object when the scanning is done

• GET /v1/status - Request kubescape scan status
• • query id=<string> -> Check status of a specific scan. If empty, it will check if any scan is still in progress

Response:

When scanning is in progress

{
  "id": <str>,                    // scan ID
  "type": "busy",                 // response object type
  "response": <message:string>    // message indicating scanning is still in process
}

When scanning is not in progress

{
  "id": <str>,                    // scan ID
  "type": "notBusy",              // response object type
  "response": <message:string>    // message indicating scanning is successfully done
}

Delete cached results

• DELETE /v1/results - Delete kubescape scan results from storage. If empty will delete the latest results
• • query id=<string>: Delete ID of specific results
• • query all: Delete all cached results

Objects

Trigger scan object

{
  "format": <str>,               // results format [default: json] (same as 'kubescape scan --format')
  "excludedNamespaces": [<str>], // list of namespaces to exclude (same as 'kubescape scan --excluded-namespaces')
  "includeNamespaces": [<str>],  // list of namespaces to include (same as 'kubescape scan --include-namespaces')
  "useCachedArtifacts"`: <bool>, // use the cached artifacts instead of downloading (offline support)
  "hostScanner": <bool>,         // deploy Kubescape host-sensor daemonset in the scanned cluster. Deleting it right after we collecting the data. Required to collect valuable data from cluster nodes for certain controls
  "keepLocal": <bool>,           // do not submit results to Kubescape cloud (same as 'kubescape scan --keep-local')
  "account": <str>,              // account ID (same as 'kubescape scan --account')
  "access-key": <str>,            // account ID (same as 'kubescape scan --accessKey')
  "targetType": <str>,           // framework/control
  "targetNames": [<str>]         // names. e.g. when targetType==framework, targetNames=["nsa", "mitre"]
}

Response object

{
  "id": <str>,                      // scan ID
  "type": <responseType:str>,       // response object type
  "response": <object:interface>    // response payload as list of bytes
}

Response object types

• "v1results" - v1 results object
• "busy" - server is busy processing previous requests
• "notBusy" - server is not busy processing previous requests
• "ready" - server is done processing request and results are ready
• "error" - error object

API Examples

Default scan

1. Trigger kubescape scan

curl --header "Content-Type: application/json" --request POST --data '{"hostScanner":true}' http://127.0.0.1:8080/v1/scan

2. Get kubescape scan results

curl --request GET http://127.0.0.1:8080/v1/results -o response.json

Trigger scan and wait for the scan to end

curl --header "Content-Type: application/json" --request POST --data '{"hostScanner":true}' http://127.0.0.1:8080/v1/scan?wait -o scan_results.json

Scan single namespace with a specific framework

curl --header "Content-Type: application/json" \
  --request POST \
  --data '{"hostScanner":true, "includeNamespaces": ["kubescape"], "targetType": "framework", "targetNames": ["nsa"] }' \
  http://127.0.0.1:8080/v1/scan

Data profiling

Analyze profiled data using pprof. How to use

example:

go tool pprof http://localhost:6060/debug/pprof/heap

Examples

• Prometheus
• Microservice

Supported environment variables

• KS_ACCOUNT: Account ID
• KS_EXCLUDE_NAMESPACES: List of namespaces to exclude, e.g. KS_EXCLUDE_NAMESPACES=kube-system,kube-public
• KS_INCLUDE_NAMESPACES: List of namespaces to include, rest of the namespaces will be ignored. e.g. KS_INCLUDE_NAMESPACES=dev,prod
• KS_HOST_SCAN_YAML: Full path to the host scanner YAML
• KS_FORMAT: Output file format. default is json
• KS_ENABLE_HOST_SCANNER: Enable the host scanner feature
• KS_DOWNLOAD_ARTIFACTS: Download the artifacts every scan
• KS_LOGGER_NAME: Set logger name
• KS_LOGGER_LEVEL: Set logger level