The Agent Skills Directory

[SAFE]: Comprehensive analysis of the instructions and supporting documentation identifies no malicious logic, credential harvesting, or unauthorized communication patterns.- [COMMAND_EXECUTION]: The skill includes a utility script scripts/email_sequence_analyzer.py used to grade email drafts. An audit confirms the script uses only safe, standard library modules and lacks capabilities for network access or persistent file system modifications.- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it processes user-provided drafts and local context files such as marketing-context.md.
Ingestion points: User-supplied email content and marketing-context.md (if present).
Boundary markers: None identified.
Capability inventory: Text processing and generation, and execution of the local analyzer script.
Sanitization: None identified. Given the skill's scope as a writing assistant and the absence of high-privilege capabilities, this surface is considered a functional requirement rather than a security threat.

cold-email