The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill retrieves and displays event logs from the Pulumi Neo API. This ingestion of external data creates a potential indirect prompt injection surface where instructions hidden in infrastructure names or Neo's output could attempt to override agent behavior.\n
Ingestion points: The get_events function in scripts/neo_task.py fetches data from the api.pulumi.com endpoint.\n
Boundary markers: External data is displayed without delimiters or specific instructions for the agent to ignore embedded commands.\n
Capability inventory: The skill has network access via the requests library and executes the pulumi CLI tool.\n
Sanitization: The script performs no filtering or sanitization of the fetched event content.\n- [COMMAND_EXECUTION]: The script scripts/neo_task.py uses subprocess.run to call pulumi org get-default. This execution uses a hardcoded list of arguments and does not interpolate user input, providing a safe method for configuration detection.

pulumi-neo