The Agent Skills Directory

[SAFE]: The skill consists entirely of textual instructions and metadata defining the agent's role, personality, and motivation. No malicious command execution, data exfiltration patterns, or obfuscation techniques were identified.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves analyzing external data such as code, architectural designs, and arguments from other agents to resolve deadlocks. This creates a functional surface for indirect prompt injection where instructions could be hidden in analyzed content. However, the skill files do not implement any unsafe tool usage or data handling that would escalate this beyond a general risk factor.
Ingestion points: Processes architectural documents, test results, codebases, and arguments from other agents.
Boundary markers: None explicitly defined in the skill instructions.
Capability inventory: No executable code or specific tool definitions are included in the provided files.
Sanitization: No sanitization logic is described in the skill definition.

decision-maker