The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads the ax CLI binary from the developer's GitHub repository (github.com/toss/apps-in-toss-ax) using curl or Invoke-WebRequest to a local cache directory.
[REMOTE_CODE_EXECUTION]: The wrapper scripts run-ax.sh and run-ax.ps1 download a binary and immediately execute it with user-provided arguments.
[COMMAND_EXECUTION]: The skill executes shell commands to manage binary versions, download archives, and extract them using the tar utility.
[INDIRECT_PROMPT_INJECTION]: The skill ingests content from external documentation sources, which introduces a potential attack surface if those sources contain malicious instructions.
Ingestion points: Documentation text and code examples retrieved via the get doc, get tds-rn, get tds-web, and get example commands.
Boundary markers: The instructions do not specify the use of delimiters or boundary markers to separate retrieved data from the agent's instructions.
Capability inventory: Execution of the ax binary via shell scripts which can interact with the file system and network.
Sanitization: There is no evidence of sanitization or filtering applied to the retrieved documentation content before it is provided to the agent.

docs-search