The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill includes a manual correction workflow where the agent reads and modifies transcribed lyrics output. This represents a surface for indirect prompt injection if the audio content or transcription service output contains malicious instructions.
Ingestion points: The agent reads the transcribed LRC file content using the Read tool as part of the 'Post-Transcription Lyrics Correction' workflow (SKILL.md).
Boundary markers: The instructions lack specific boundary markers or 'ignore embedded instructions' warnings for the agent when processing the untrusted transcription data.
Capability inventory: The skill uses the Bash tool for shell commands and the Write tool for file system modifications.
Sanitization: No sanitization or validation is applied to the transcription text before it is presented to the agent for correction.
[DYNAMIC_EXECUTION]: The utility script scripts/acestep-lyrics-transcription.sh executes embedded Python code using python -c to perform LRC and SRT format conversions. While these are static templates within the script, they represent dynamic execution of logic at runtime based on external API data.
[EXTERNAL_DOWNLOADS]: The skill makes network requests to well-known AI services, specifically OpenAI (api.openai.com) and ElevenLabs (api.elevenlabs.io), to perform audio transcription. These operations are essential to the skill's primary purpose and target established service providers.

acestep-lyrics-transcription