The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to facilitate a 'copy to clipboard' feature using OS-specific commands such as pbcopy, xclip, or clip. The content passed to these commands is the 'optimized prompt,' which is directly derived from untrusted user input. This creates a potential command injection vector if the agent interpolates the prompt into the shell command without proper escaping (e.g., if a prompt contains backticks or shell meta-characters).
[INDIRECT_PROMPT_INJECTION]: The skill's primary purpose is to process untrusted data (user-supplied drafts) to provide optimization services, which establishes an attack surface for indirect prompt injection.
Ingestion points: Untrusted user data enters the agent context through the initial request and during the 'Intake & Assessment' phase (Step 1) in SKILL.md.
Boundary markers: The skill does not implement robust structural boundary markers (such as randomized XML tags or strict schema validation) to isolate user-provided text from the skill's operational instructions.
Capability inventory: The skill has the ability to execute shell commands (Bash), write to the local file system (Write), and read local files (Read).
Sanitization: The instructions include behavioral guidance for the agent to 'consider injection attacks' and 'use delimiters,' but it lacks technical enforcement mechanisms to prevent malicious user-provided content from influencing the agent's behavior during the optimization process.

accelint-prompt-manager