The Agent Skills Directory

REMOTE_CODE_EXECUTION (HIGH): The skill performs unverified remote code execution by downloading a shell script and piping it directly to the bash interpreter.
Evidence: curl -fsSL https://raw.githubusercontent.com/diverta/kuroco-skills/main/skills/kuroco-docs/scripts/sync-docs.sh | bash -s -- "${CLAUDE_PLUGIN_ROOT}" found in SKILL.md.
Context: While the script is intended for syncing Kuroco documentation, the repository diverta/kuroco-skills is not a trusted source, meaning the script could be maliciously modified without the user's knowledge.
EXTERNAL_DOWNLOADS (MEDIUM): The skill downloads both scripts and data files from an external source not identified as a trusted organization.
COMMAND_EXECUTION (LOW): The skill uses various shell commands (ls, cat, date, grep) to manage local files and search documentation.
PROMPT_INJECTION (LOW): The skill is vulnerable to indirect prompt injection because it ingests and processes external data (markdown files) that could contain malicious instructions.
Ingestion points: External markdown files downloaded to ${CLAUDE_PLUGIN_ROOT}/docs/.
Boundary markers: Absent; documentation is read and displayed directly without delimiters or instruction-ignore warnings.
Capability inventory: The skill possesses file system access, command execution (bash), and network access (curl).
Sanitization: Absent; there is no validation or sanitization of the content within the downloaded documentation before it is processed by the agent.

kuroco-docs