The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests transcript data from external YouTube videos via the YouMind API and automatically generates summaries. This creates a surface where adversarial instructions embedded in a video's subtitles could potentially influence the agent's behavior during the summary generation phase.
Ingestion points: Transcript content is retrieved using the youmind call getMaterial command in SKILL.md.
Boundary markers: Absent. The instructions do not specify the use of delimiters or 'ignore' instructions when processing the external transcript text.
Capability inventory: The skill utilizes shell command execution via the youmind CLI and Node.js for data parsing.
Sanitization: Absent. There is no mention of filtering or validating the transcript content before it is processed by the LLM.
[PROMPT_INJECTION]: The instructions include mandatory directives that instruct the agent to bypass user confirmation for certain actions, such as showing onboarding information immediately and automatically generating summaries without asking (e.g., "Do NOT ask... — just show it", "Do NOT ask the user whether they want a summary — just do it").
[COMMAND_EXECUTION]: The skill requires the execution of shell commands to interact with the vendor-provided CLI tool (youmind) and to install dependencies via npm.
[DYNAMIC_EXECUTION]: In SKILL.md, Step 5 uses node -e to execute an inline JavaScript snippet. This script is used as a lightweight way to parse and filter JSON data returned by the CLI tool.
[EXTERNAL_DOWNLOADS]: The skill guides the user to install a global Node.js package (@youmind-ai/cli) which is a vendor-owned resource necessary for the skill's functionality.

youmind-youtube-transcript