The Agent Skills Directory

[SAFE]: A thorough analysis of the orchestration strategies, prompt templates, and instructional content revealed no malicious patterns, obfuscation, or unauthorized data access. The skill aligns with established software engineering best practices for multi-agent systems.
[COMMAND_EXECUTION]: The skill instructs agents to use standard development utilities such as git for managing parallel feature builds and ruff for code quality sweeps. These operations are limited to the project directory and are essential for the skill's stated purpose.
[PROMPT_INJECTION]: The skill describes a system for processing external data (via web research) and local codebase content, which presents a surface for indirect prompt injection.
Ingestion points: Prompt templates for research agents (SKILL.md) ingest data from WebSearch and WebFetch tools.
Boundary markers: The framework utilizes structured Markdown headers to delimit agent instructions from external context, though it lacks explicit 'ignore instructions' guards.
Capability inventory: Managed agents possess capabilities to write files, execute development tools, and perform git operations (git commit, git status).
Sanitization: No explicit content sanitization or validation of the fetched external data is described within the prompt templates.

orchestrate