The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool and accepts user-provided input through the $ARGUMENTS variable (album name). While instructions guide the agent to perform specific tool calls, the combination of raw user input and shell access creates a potential command injection surface if the agent does not properly sanitize the input before execution.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill ingests untrusted data from various local files including RESEARCH.md, SOURCES.md, and individual track files using Read, Grep, and Glob tools.
Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from following malicious instructions that could be embedded within the source citations or research notes being verified.
Capability inventory: The agent has permissions to modify the file system via the Edit tool and execute system-level commands through Bash and the bitwize-music-mcp extension.
Sanitization: The skill lacks explicit sanitization or filtering logic for the content it extracts from research files, relying instead on human oversight during the verification phase.

verify-sources