The Agent Skills Directory

[COMMAND_EXECUTION]: The skill directs the agent to execute a local command-line interface tool named bili-transcript. It uses subcommands like doctor, transcribe, and batch to perform its primary functions. These commands are used to interact with the local system's shell to process media files.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process content from external Bilibili URLs, which introduces a surface for indirect prompt injection if the video content or metadata contains malicious instructions intended for the AI.
Ingestion points: Public Bilibili video metadata and audio streams accessed via the CLI tool (SKILL.md).
Boundary markers: The instructions do not specify any explicit delimiters or 'ignore' instructions to prevent the agent from following directions that might be present in the transcribed text.
Capability inventory: The skill possesses the capability to execute shell commands (bili-transcript) and perform file system writes to output directories (SKILL.md).
Sanitization: The skill does not mention any sanitization or filtering of the machine-generated transcripts before they are delivered to the user or agent context.

bilibili-transcript