The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides explicit instructions for the AI agent to execute host binaries (e.g., stata-mp) via the command line to run analysis scripts.
[COMMAND_EXECUTION]: Documentation is provided for using Stata's native shell and ! commands, which allow for the execution of arbitrary operating system commands from within a Stata session.
[EXTERNAL_DOWNLOADS]: The skill details methods for downloading and installing external software packages from the Statistical Software Components (SSC) archive, the Stata Journal, and arbitrary GitHub repositories.
[EXTERNAL_DOWNLOADS]: Instructions are included for importing data directly from remote URLs using commands like import delimited and copy.
[DYNAMIC_EXECUTION]: Extensive documentation is provided for integrating Stata with Python (python:) and R (rcall), enabling the execution of code in those languages and data exchange between environments.
[REMOTE_CODE_EXECUTION]: The skill facilitates remote code execution by teaching the agent how to run Stata scripts that can, in turn, trigger system-level commands or scripts in other languages.
[INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection as it is designed to ingest and process data from untrusted external sources (CSV, Excel, etc.) without specifying boundary markers or sanitization procedures.
Ingestion points: references/data-import-export.md (documented import delimited and import excel commands).
Boundary markers: Absent; there are no instructions to use delimiters or "ignore embedded instructions" warnings when processing data.
Capability inventory: SKILL.md (shell execution of stata binary), references/external-tools-integration.md (shell, !, python:, rcall:), packages/package-management.md (ssc install, github install).
Sanitization: Absent; no filtering or escaping of external content is mentioned.

stata