The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes ls ~/.claude-context-optimizer/templates/ and a Node.js script via the shell. While these are for local functionality, executing shell commands with directory variables ($(pwd)) is a pattern that can be sensitive to malicious path names containing shell metacharacters.
[DYNAMIC_EXECUTION]: The skill invokes a Node.js script located at ${CLAUDE_PLUGIN_ROOT}/src/tracker.js. This executes logic defined within the skill's own package to generate file suggestions.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from the local file system, including templates and historical patterns. This creates a surface where malicious content within those local files or directory names could influence the agent's behavior.
Ingestion points: Data enters the context from the ~/.claude-context-optimizer/templates/ directory and the output of the tracker.js script.
Boundary markers: None are explicitly mentioned to separate the suggestions from system instructions.
Capability inventory: The skill has the ability to execute shell commands (ls, node).
Sanitization: No specific sanitization or validation of the historical data or template content is defined in the instructions.

smart-loader