The Agent Skills Directory

[PROMPT_INJECTION] (LOW): The skill is vulnerable to indirect prompt injection due to its core workflow of reading and summarizing project documentation that may contain untrusted content. • Ingestion points: CONTEXT.md, ROADMAP.md, and Git logs/status. • Boundary markers: Absent (no clear delimiters or 'ignore' instructions for processed content). • Capability inventory: Git operations, file writes to docs/, and execution of external tools like Gemini CLI. • Sanitization: None identified in the provided skill definitions.
[COMMAND_EXECUTION] (LOW): The skill metadata and documentation describe the use of shell commands and external CLI tools. While these are used for development purposes (setup, linting, and AI-assisted review), they constitute a capability surface that could be leveraged if the agent is misled by malicious content in ingested files.

session-support