Ensures email recipients, content, tone, attachments, claims, and approvals are verified and safe before sending or scheduling messages.
Security Analysis
medium confidenceThis is an instruction-only email safety checklist that requires recipient, privacy, attachment, claim, tone, and explicit approval checks before email is sent.
The artifacts consistently describe an email-send guardrail and state that the skill does not send email, run code, install packages, or persist data by itself.
The instructions are restrictive rather than expansive: they require explicit approval, recipient verification, redaction, attachment review, and blocking unsafe sends.
There is no install spec, code, dependency installation, or command execution. Registry provenance is limited because the source is unknown and no homepage is provided.
The skill may use an optional configured checker or manual review path for redacted email-review payloads; this is purpose-aligned but should remain limited to trusted configured interfaces.
The manifest declares no credential requirements, no file writes, no command execution, and no memory persistence.
Guidance
This skill appears safe and purpose-aligned as an email pre-send guardrail. Before installing, note that the source metadata is limited, and if you connect an AANA checker, it should be a trusted/admin-approved interface that receives only minimal redacted review data.
Latest Release
v1.0.0
Initial release of the AANA Email Send Guardrail Skill. - Provides detailed instructions for agents to verify recipients, content, tone, private data, attachments, and claims before sending email or related messages. - Separates drafting, revising, approval, and send actions to prevent accidental or unauthorized emails. - Lists explicit pre-send checks, privacy, recipient, attachment, claim, tone, and approval rules. - Introduces the AANA Email Gate Loop for comprehensive, step-by-step email risk assessment. - Defines a structured output pattern for email status classification and decision-making.
More by @mindbomber
Published by @mindbomber on ClawHub
