The Agent Skills Directory

[Prompt Injection] (HIGH): The skill explicitly commands the agent to override standard safety guardrails and interactive consent layers. Instructions such as 'IMMEDIATELY run the video analyzer
do not ask for permission' and 'DO NOT ask permission
run analysis immediately' are direct attempts to bypass the agent's typical request-for-approval workflow for tool execution.
[Command Execution] (MEDIUM): The skill triggers the execution of a local script (tools/video-analyzer.py) using a user-provided string (YouTube URL) as a direct argument. This creates a command injection risk if the URL is not strictly validated before being passed to the shell.
[Indirect Prompt Injection] (LOW): The skill exhibits an indirect prompt injection surface by ingesting and acting upon untrusted data from YouTube transcripts.
Ingestion points: YouTube transcripts and metadata processed via the video-analyzer.py tool.
Boundary markers: Absent; the instructions do not specify any delimiters or warnings to ignore malicious instructions embedded within the transcripts.
Capability inventory: Subprocess execution via python and file writing to projects/analysis/.
Sanitization: None mentioned; the skill assumes the transcript content is benign and proceeds directly to implementation planning based on the extracted concepts.

pine-visualizer