The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes existing agent definition files to refine or evolve them. This represents a potential surface for indirect prompt injection if the source files contain malicious instructions.\n- Ingestion points: The agent reads existing Markdown files (e.g., agents/<agent-name>.md) using cat and head commands during the assessment and validation phases.\n- Boundary markers: There are no explicit instructions or delimiters used to tell the agent to ignore potentially malicious embedded content within the source agent files during evolution.\n- Capability inventory: The skill has access to Bash, Edit, and Write tools, and it executes several project-specific Node.js and shell scripts.\n- Sanitization: The skill does not perform validation or sanitization of the content within the agent files specifically to prevent instruction injection before processing.

evolve-agent