The Agent Skills Directory

[COMMAND_EXECUTION]: In scripts/command_center.py, the script utilizes subprocess.check_output to run git rev-parse --show-toplevel. This is a common and safe method to locate the project root directory, as the command uses hardcoded arguments and does not process external or user-controlled input.
[INDIRECT_PROMPT_INJECTION]: The skill aggregates task descriptions and error logs into a markdown file, which creates a potential surface for indirect prompt injection if the agent interprets the dashboard content as instructions.
Ingestion points: The skill reads fields such as description and reason from .agent/state/tasks.json, .agent/state/last-run.json, and .agent/state/errors.json in scripts/command_center.py.
Boundary markers: The generated status.md dashboard does not include delimiters or instructions for the agent to ignore embedded instructions within the task logs.
Capability inventory: The skill's capabilities are restricted to executing a specific git command and writing the resulting dashboard to the local file system.
Sanitization: The script performs no escaping or validation of the data retrieved from the state files before formatting it into the markdown dashboard.

command-center