The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-supplied lyrics and incorporates them as arguments for local diagnostic scripts, creating a theoretical attack surface for indirect injection or command manipulation.
Ingestion points: User-provided lyrics submitted to the agent for diagnostic analysis in SKILL.md.
Boundary markers: Absent; the instructions do not define specific delimiters or warnings to ignore instructions embedded within the lyrics being analyzed.
Capability inventory: The skill utilizes the bun runtime to execute three local scripts: cliche-check.ts, meter-check.ts, and rhyme-check.ts across its diagnostic states.
Sanitization: Absent; the skill does not provide instructions to escape or validate the input text before passing it to the script execution layer.
[COMMAND_EXECUTION]: The skill relies on the execution of local TypeScript scripts to perform its core analytical functions. The SKILL.md instructions guide the agent to run these scripts (e.g., bun run scripts/cliche-check.ts "lyrics") to generate findings for the user. While this is the intended primary purpose, it involves subprocess spawning with user-controlled data.

lyric-diagnostic