The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute local CLI commands (e.g., memory-hub catalog-repair, memory-hub discover) and perform filesystem inspections within the .memory/ directory. These actions are restricted to maintenance and diagnostic purposes as defined in the skill scope.
[PROMPT_INJECTION]: The skill processes session transcripts through the memory-hub session-extract command, which introduces a surface for indirect prompt injection from untrusted conversation data.
Ingestion points: External data enters the system through session transcript files specified in the --file argument.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands within the transcripts are provided in the skill text.
Capability inventory: The skill possesses the capability to execute shell commands via the Bash tool and perform read operations on the project directory.
Sanitization: The documentation does not specify content sanitization or validation for the processed transcripts, relying on the CLI tool's internal classification logic.

memory-admin