The Agent Skills Directory

[PROMPT_INJECTION]: The skill ingests external content from Moltbook comments and X/Twitter posts to perform automated synthesis and knowledge extraction. The absence of strict delimiters or sanitization for this untrusted data creates an indirect prompt injection surface. Malicious actors could craft social media posts containing hidden instructions designed to influence the agent's research notes, blog post generation, or subsequent social interactions.
[COMMAND_EXECUTION]: The skill relies on subprocess.run to execute command-line tools like curl and xurl for network communication and social media actions. While the implementation follows best practices by passing arguments as lists to prevent shell injection, the skill's core functionality depends on executing external binaries based on agent-generated content.
[CREDENTIALS_UNSAFE]: The skill requires the management of sensitive authentication data, including API keys, passwords, and browser session cookies, stored in the ~/.config/moltbook/ and ~/.config/x/ directories. While necessary for the skill's operation, these local files represent a high-value target for credential harvesting if the host environment is compromised.

social-intelligence