The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to build and execute shell commands using user-provided task prompts as command-line arguments (e.g., bun src/index.ts subagent run "task-1" "промпт задачи 1"). This pattern allows for command injection if the task descriptions contain shell metacharacters like semicolons, backticks, or redirection operators.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes external task descriptions and integrates them into active command execution without sanitization.
Ingestion points: Task prompts are accepted from user input and passed to the bun runtime in the shell scripts described in SKILL.md.
Boundary markers: Absent. The task input is interpolated directly into the bash command string without delimiters or instructions for the agent to ignore embedded commands.
Capability inventory: The skill utilizes shell execution (bash), background process management (&), file system operations (ls, tee, wc), and the claudeclaw progress tool.
Sanitization: Absent. There is no evidence of validation, escaping, or filtering of the user-provided strings before they are executed as part of a command line.

subagent-coordinator