The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains a 'Mode Detection Algorithm' that analyzes user input for keywords like 'overwhelmed' or 'adhd' to switch between base and accessibility modes. This logic relies on interpreting user-provided text to influence agent behavior.
[PROMPT_INJECTION]: Indirect Prompt Injection vulnerability surface.
Ingestion points: The skill ingests the user's message to evaluate distress signals and mode requests (SKILL.md, 'Mode Detection Algorithm' section).
Boundary markers: No specific boundary markers or 'ignore embedded instructions' warnings are present to separate user content from the agent's logic.
Capability inventory: The skill possesses broad capabilities including Bash(cmd:*), Write, Edit, Grep, and Glob (SKILL.md frontmatter).
Sanitization: No sanitization or validation of user-provided keywords is described before they are processed by the logic used to determine the agent's behavioral constraints.
[COMMAND_EXECUTION]: The skill requests permission for the Bash(cmd:*) tool. Although the instructions focus on text and diagram generation, this grants the agent unrestricted shell access which could be exploited if the agent is misled by malicious input.
[EXTERNAL_DOWNLOADS]: The skill generates links to mermaid.live, a well-known and trusted external service for rendering and editing Mermaid diagrams. These links include encoded diagram data to allow the user to visualize their organizational tools.

neurodivergent-visual-org