The Agent Skills Directory

[SAFE]: The skill defines a set of roles and workflows for collaborative software design brainstorming. It utilizes the Task tool to coordinate subagents that perform codebase analysis, trade-off evaluation, and risk assessment.
[SAFE]: External references and links point to well-known educational, professional, and technical resources (e.g., IEEE, CMU SEI, Medium, Anthropic). No downloads from untrusted or malicious sources are performed.
[PROMPT_INJECTION]: The skill processes external data (user arguments, codebase content, and web search results) which creates a surface for indirect prompt injection. However, this is inherent to its primary purpose as an analysis tool and is considered a low risk. The orchestration follows best practices for multi-agent synthesis.
[PROMPT_INJECTION]: Ingestion points: Processes user-provided arguments in SKILL.md, codebase content via the Explorer agent, and external web search results via the Scout agent.
[PROMPT_INJECTION]: Boundary markers: Lacks explicit delimiters or instructions to ignore embedded commands within ingested data in the subagent templates in references/agent-roles.md.
[PROMPT_INJECTION]: Capability inventory: Utilizes the Task tool to spawn subagents, reads codebase files, and performs web searches.
[PROMPT_INJECTION]: Sanitization: No input sanitization or validation logic is present for the processed external data.

brainstorm