The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill encourages the agent to 'act by default' and perform 'silent' decisions based on facts retrieved from external sources. This creates a surface for indirect prompt injection where malicious instructions embedded in web pages or documents could influence the agent's behavior without the user's knowledge.
Ingestion points: The SKILL.md file specifies that the agent should gather facts from 'files, docs, standards, and web' to inform its decisions.
Boundary markers: There are no instructions provided to use delimiters or ignore potential commands found within the data retrieved from external sources.
Capability inventory: The skill allows the agent to autonomously perform actions such as modifying user configuration files, changing schemas, and acting on taste or implementation choices.
Sanitization: The skill does not describe any validation or sanitization process for the data it ingests before using that data to justify its autonomous actions.

discretion