The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute a command: memory-hub index <bucket> <file> --topic <name> --summary "<description>" [--anchor <anchor>]. The direct interpolation of variables into a shell string is a known risk for arbitrary command execution if the inputs are not strictly validated.
[PROMPT_INJECTION]: This skill presents an indirect prompt injection surface as it processes untrusted data to modify the system index.
Ingestion points: Parameters bucket, file, topic, summary, and anchor are provided during execution, potentially originating from untrusted user content.
Boundary markers: No explicit boundary markers or "ignore instructions" warnings are present in the command template.
Capability inventory: The skill has Bash and Write capabilities, allowing it to execute system commands and modify files in the .memory/ directory.
Sanitization: There is no evidence of escaping or sanitization for shell metacharacters (e.g., ;, &, |) in the input fields before they are passed to the bash shell.

memory-index