The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on executing local helper scripts, specifically scripts/render.sh for HTML rendering and scripts/degrade-rich.mjs for transforming markdown structures. These are invoked by the agent to perform the skill's core functions.
[EXTERNAL_DOWNLOADS]: The rendering script scripts/render.sh uses npx --yes markmap-cli to execute the mindmap generator. This command may download the markmap-cli package from the public npm registry if it is not already present in the environment.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from various sources, creating an opening for indirect prompt injection attacks.
Ingestion points: The skill retrieves content from user-specified URLs and reads from local file paths as defined in the SKILL.md workflow.
Boundary markers: The skill uses simple text headers such as SOURCE CONTENT: in references/judge-panel.md to separate external data from its instructions, which is a weak boundary that can be bypassed by instructions hidden within the source data.
Capability inventory: The skill is granted access to read and write files, fetch web content, and execute shell commands, providing a significant set of actions that could be abused if an injection is successful.
Sanitization: The provided instructions and scripts do not implement sanitization, filtering, or escaping for the ingested content before it is passed to the agent or its subagents.

mindmap