The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill directs users and the agent to an external 'registre des outils' (tool registry) and multiple platform integration guides (Klaviyo, Brevo, Twilio, etc.) hosted on GitHub at https://github.com/mbzadev/skills/. The stated author handle is mabzadev. The handle used in the URLs (mbzadev) is a variation missing a character, which corresponds to established patterns for typosquatting or impersonating vendor resources.
[REMOTE_CODE_EXECUTION]: The skill encourages the agent to consult the external tool registry for implementation ('mise en œuvre') and explicitly notes MCP (Model Context Protocol) support for several tools. This creates a pathway for the agent to fetch and potentially execute tool definitions or configurations from a potentially malicious impersonator account.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to its reliance on external product context files.
Ingestion points: Instructions specify reading product-marketing context from .agents/product-marketing.md, .claude/product-marketing.md, and product-marketing-context.md.
Boundary markers: None. The skill does not instruct the agent to use delimiters or ignore instructions embedded within the ingested files.
Capability inventory: The skill includes tool integrations and access to a tool registry to execute marketing campaigns.
Sanitization: Absent. Content from the marketing context files is interpolated directly into the agent's strategy and SMS sequence design without filtering.

sms