twitter-automation
Pass
Audited by Gen Agent Trust Hub on Nov 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references installation scripts and documentation from the author's official GitHub repository (
inference-sh/skills) and website (inference.sh). These are standard operations for setting up the required environment and are considered safe vendor resources. - [COMMAND_EXECUTION]: The skill uses the
beltCLI tool to perform Twitter/X operations. The execution environment is appropriately restricted via theallowed-toolsconfiguration, ensuring only the intended tool can be invoked. - [INDIRECT_PROMPT_INJECTION]: The skill contains a vulnerability surface for indirect prompt injection as it facilitates posting content to social media that may be derived from external sources or other AI tools.
- Ingestion points: Data enters the context via the
--inputparameter inbelt app runcommands found inSKILL.md. - Boundary markers: The skill uses structured JSON to delimit input data, which provides a level of structural boundary against instruction leakage.
- Capability inventory: The skill can post, like, retweet, delete, and send direct messages on Twitter/X.
- Sanitization: No explicit sanitization or filtering of input text is demonstrated in the examples; the skill relies on the underlying platform's safety layers and the agent's own constraints.
Audit Metadata