The Agent Skills Directory

[OBFUSCATION]: The script scripts/check-connection.mjs uses a Unicode escape sequence (\u002d) to represent the hyphen in the x-api-key header name.
[OBFUSCATION]: The verification script dynamically constructs API key prefixes (sk- and proj-) using string joining (join("-")) rather than static literals, which avoids simple pattern-based detection of secret-handling code.
[OBFUSCATION]: The script accesses the fetch function using bracket notation on the global object (globalThis["fetch"]), an indirect method of invoking network operations.
[DATA_EXFILTRATION]: The skill's verification script reads the sensitive MERMAIL_API_KEY from the process environment and transmits it to an external endpoint (console.mermail.app) via a POST request.
[INDIRECT_PROMPT_INJECTION]: The skill identifies an attack surface for indirect prompt injection as it processes external data from tool results and server responses.
Ingestion points: Tool results, server error messages, and mailbox/email data are ingested into the agent context as described in SKILL.md and references/security.md.
Boundary markers: No technical delimiters or structural markers are implemented to isolate this external data within the agent's prompts.
Capability inventory: The skill includes network access via fetch and facilitates access to email management, workspace administration, and mailbox provisioning tools.
Sanitization: The skill relies on instructional guidelines to the agent (e.g., redacting credentials and treating data as untrusted) rather than technical sanitization or filtering of the ingested content.

mermail-mcp