Mixpanel

Query Mixpanel analytics with funnels, retention, segmentation, and event tracking via REST API.

This skill appears coherent for querying Mixpanel, but before installing consider the following: 1) The skill requires a Mixpanel service account and secret — treat MP_SERVICE_SECRET as sensitive since it can enable queries and exports; create a least-privilege service account and rotate credentials regularly. 2) The SKILL.md example uses the base64 utility but only lists curl and jq as required binaries — ensure base64 is available in your agent environment or add it to required tools. 3) Cached query results are stored in ~/mixpanel/ — if those results contain PII or sensitive analytics, restrict directory permissions and consider where backups or sync tools might copy that data. 4) The skill uses the raw export endpoint which can return detailed event data; only enable this capability for accounts that are allowed to export raw event data. 5) The skill claims it will not send data outside Mixpanel; confirm the agent runtime you use executes those curl commands locally (and does not proxy them through an external host you do not control). If any of these points are unacceptable (for example, service account secrets you cannot provide or strict data-export restrictions), do not install or limit the service-account permissions before enabling the skill.

Type: OpenClaw Skill Name: mixpanel Version: 1.0.1 The OpenClaw Mixpanel skill is benign. It clearly defines its purpose of interacting with the Mixpanel API for analytics, using standard tools like `curl` and `jq`. It correctly handles sensitive credentials by requiring them as environment variables (`MP_SERVICE_ACCOUNT`, `MP_SERVICE_SECRET`, `MP_PROJECT_ID`) and explicitly instructs the agent not to store them in files or ask users to paste them in chat (`SKILL.md`, `memory-template.md`, `setup.md`). All network calls are directed to legitimate Mixpanel domains, and the skill explicitly disavows sending data to other services. There is no evidence of data exfiltration to unauthorized endpoints, malicious execution, persistence mechanisms, or prompt injection aiming for harmful actions.