The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands (mkdir -p and git mv) to reorganize the file system. These operations are scoped to the docs/threads/ directory within the repository.
[PROMPT_INJECTION]: The skill identifies an indirect prompt injection surface through the processing of untrusted data:
Ingestion points: The agent reads file content (titles and headers) from subdirectories like .pending-decisions/ and .pending-reviews/ within the target thread folder.
Boundary markers: No specific delimiters or "ignore" instructions are used when interpolating these headers into the prompt for user confirmation.
Capability inventory: The agent has the ability to execute shell commands (git mv, mkdir) and perform file reads.
Sanitization: No sanitization or escaping is performed on the headers read from the files before they are presented to the user.
[SAFE]: The skill's metadata and author information are consistent with the provided context, and no network operations or access to sensitive system credentials were detected.

archive-thread