The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands to manage its file-based outputs. It executes mkdir to create directories, cat to write tracking documentation, and source to load local shell scripts (id-generator.sh and index-updater.sh) from the plugin's library directory. It also has explicitly granted permission to edit the settings.yaml configuration file.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it reads and parses project requirement documentation (PRD) to suggest tracking points. Ingestion points: Step 1 (Mode A) reads PRD files or text from $JAAN_OUTPUTS_DIR. Boundary markers: No specific delimiters or safety instructions are defined to separate the ingested PRD content from the agent's instructions. Capability inventory: The skill has permission to Write to the output directory, Edit configuration files, and execute shell commands. Sanitization: No validation or sanitization logic is described for the content extracted from the PRD files.

data-gtm-datalayer