Security audit
Security checks for vulnerabilities and agentic risk
This is a straightforward email API skill, but it sends message data through AgentLair and should only be used for user-approved email activity.
Install only if you want agents to use AgentLair for email. Treat AGENTLAIR_API_KEY as a secret, review recipients and message content before sending, and avoid regulated or sensitive information unless your organization has approved AgentLair and Amazon SES as email processors.
The manifest description says 'no credentials to configure', which contradicts the actual usage documented elsewhere in the file. The skill explicitly requires an API key in the Authorization header and names AGENTLAIR_API_KEY as the primary environment variable, so authentication credentials are in fact required.
This skill is explicitly designed for external transmission: it sends email, reads inbox contents, and uses a remote API authenticated by an API key. In context, the behavior is expected, but it still creates a real data exfiltration channel because message bodies, addresses, and metadata leave the local environment and are processed by external infrastructure.
primaryEnv: AGENTLAIR_API_KEY
requires:
bins:
- curl
tags:
- email
- agent-identity
The skill instructs agents to send email and retrieve inbox contents from an external service without prominently warning about disclosure of message content, recipient data, and metadata to a third party. In an agent setting, this can cause unintentional transmission of sensitive user or enterprise information outside approved channels.
The example session normalizes sending email to an external recipient on the user's behalf without requiring explicit confirmation of recipient, sender identity, and message contents. This can enable impersonation, accidental outreach, or unauthorized external communications by an autonomous agent.
The example workflow includes live calls to an external service to create a key, claim an address, and send mail, which establishes a concrete outbound communications path. In an agent ecosystem, such examples can be directly operationalized and used to move data or contact third parties without sufficient governance.
curl -s -X POST https://agentlair.dev/v1/auth/keys -H "Content-Type: application/json" -d '{}'
This sentence actively suggests the service operates without credentials, but the file immediately documents obtaining, storing, and presenting a bearer API key. That is an intent-level contradiction in the documentation rather than a mere omission.
No suspicious patterns detected.