The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local PowerShell script (scripts/meta-edit.ps1) using powershell.exe to perform metadata modifications. It passes several user-controlled parameters, including -ObjectPath, -Operation, and -Value, which are processed by the shell environment. This pattern requires robust argument escaping to prevent potential command injection.
[INDIRECT_PROMPT_INJECTION]: This skill has an attack surface for indirect prompt injection as it processes external, potentially untrusted data.
Ingestion points: Reads 1C metadata XML files and JSON operation definitions from the local file system via the -ObjectPath and -DefinitionFile arguments.
Boundary markers: The instructions do not define explicit boundary markers or delimiters to isolate processed data from the agent's instructions.
Capability inventory: The skill has access to powerful tools including Bash, Read, Write, and Glob, and can execute arbitrary shell commands via the powershell.exe invocation.
Sanitization: There is no explicit evidence of sanitization or validation of the content within the XML/JSON files before they are processed or used in downstream tasks like /meta-validate.
[DYNAMIC_EXECUTION]: Static analysis indicates that the supporting script scripts/meta-edit.py utilizes subprocess.run(). This indicates that the skill dynamically constructs and executes system commands based on the properties of the metadata objects being edited, which is a common but sensitive operation in development tools.

meta-edit