The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute a local bundled script, state-commit, which manages concurrent updates to the work item state file .agent-atelier/work-items.json.
[PROMPT_INJECTION]: The skill has a potential surface for indirect prompt injection as it reads and displays work item data that could contain malicious instructions.
Ingestion points: Reads data from .agent-atelier/work-items.json in the list and show subcommands.
Boundary markers: The data is presented to the agent and user within formatted markdown tables or JSON blocks.
Capability inventory: Uses Bash, Read, and Glob tools to interact with the file system.
Sanitization: No specific content filtering or sanitization of work item fields (such as titles or descriptions) is mentioned.

wi