The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external URLs and local files to generate mindmaps.
Ingestion points: Step 1 in SKILL.md handles URL and file path inputs.
Boundary markers: The instructions lack explicit delimiters or warnings for the agent to ignore instructions embedded in the external content.
Capability inventory: The agent can read and write files and execute shell commands.
Sanitization: The skill instructs the agent to compress nodes into short phrases, providing a level of natural language filtering but not full protection.
[COMMAND_EXECUTION]: The skill executes a local bash script (scripts/render.sh) to convert Markdown files to HTML.
[REMOTE_CODE_EXECUTION]: The render.sh script executes npx --yes markmap-cli, which downloads and runs a package from the NPM registry at runtime.
[EXTERNAL_DOWNLOADS]: The skill fetches content from user-provided URLs to serve as the source for mindmap generation.
[DYNAMIC_EXECUTION]: The skill uses npx to dynamically load and execute an external CLI tool during the rendering process.

mindmap-zh