The Agent Skills Directory

[SAFE]: The skill definition contains no evidence of malicious behavior, obfuscation, or security guideline bypasses. The persona is designed for strategic product marketing tasks and uses tools appropriate for that role.
[COMMAND_EXECUTION]: The Bash tool is enabled for the agent, providing command-line capabilities. Analysis confirms these are intended for legitimate file operations and task management within the local environment, with no pre-configured malicious payloads.
[EXTERNAL_DOWNLOADS]: The skill uses the WebSearch tool to retrieve external information. This is a standard research capability for the persona, and there are no instructions to download or execute external scripts or untrusted packages.
[PROMPT_INJECTION]: The skill maintains a surface for indirect prompt injection through data ingestion tools, though no active exploitation was found.
Ingestion points: External data enters the context through WebSearch and the /feedback-recall command (SKILL.md).
Boundary markers: No specific delimiters or boundary markers are defined to isolate untrusted external content.
Capability inventory: The agent has the ability to write/edit files, execute shell commands via Bash, and spawn sub-agents via Task.
Sanitization: The skill does not implement specific sanitization or validation routines for the data retrieved from external sources.

director-product-marketing