The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill documentation provides an installation command that pipes a remote script directly into the shell (curl -fsSL https://cli.inference.sh | sh). This pattern bypasses standard package management and poses a risk if the remote source or transmission is compromised.
[EXTERNAL_DOWNLOADS]: The installation script fetches pre-compiled binaries from dist.inference.sh. While the skill author provides SHA-256 checksums for manual verification, the automated installation process introduces external binary dependencies from vendor-controlled infrastructure.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8) because it processes untrusted data.
Ingestion points: User-provided content (e.g., tweet text, DM content) is passed directly into the --input argument of various infsh sub-commands.
Boundary markers: There are no boundary markers or delimiters shown in the examples to isolate user input from the command structure or to instruct the agent to ignore embedded instructions.
Capability inventory: The infsh tool has extensive capabilities, including network access to the Twitter/X API and the ability to read local files (e.g., infsh app run ... --input input.json).
Sanitization: The skill lacks visible sanitization, escaping, or validation logic for data interpolated into the automation commands.
[COMMAND_EXECUTION]: The skill utilizes the Bash(infsh *) capability. This allows the agent to execute any sub-command within the tool's ecosystem, enabling complex interactions with external APIs and potentially the local file system through JSON input files.

twitter-automation