The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands such as mkdir -p <目標專案>/.claude/handoffs and cat to verify file contents. This pattern presents a potential risk of path traversal or command injection if the user-supplied project path contains malicious sequences (e.g., ../../ or shell metacharacters) and is not properly validated by the execution environment.
[PROMPT_INJECTION]: The skill is designed to bridge context between independent agent sessions by storing information in handoff cards. This creates an attack surface for indirect prompt injection.
Ingestion points: The skill ingests user-provided task descriptions and context in SKILL.md to populate the handoff card.
Boundary markers: No specific boundary markers or instructions to ignore embedded commands are defined for the card's content.
Capability inventory: The skill utilizes directory creation (mkdir), file writing, and file reading (cat) as outlined in SKILL.md.
Sanitization: There is no evidence of sanitization, escaping, or validation of the content before it is written to the handoff file, which will be processed by a future agent session.

dropoff