The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's analysis process in reference.md instructs the agent to execute the git diff command to compare code versions and identify changes.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill ingests untrusted data from external sources, specifically git diff outputs, file lists, and ADVPL/TLPP source code provided by the user for analysis.
Boundary markers: The instructions do not define specific delimiters or "ignore" directives to prevent the agent from following malicious instructions that might be embedded within code comments or diff content.
Capability inventory: The skill utilizes file system read access and shell command execution (git diff).
Sanitization: The process lacks a validation or sanitization step for the content extracted from the files being analyzed before it is interpreted by the agent.

changelog-patterns