The Agent Skills Directory

[SAFE]: The skill specifies a fork context in its frontmatter, ensuring that the execution of the validation workflow occurs in an isolated sub-agent environment, which prevents investigation details or intermediate steps from affecting the main conversation security context.- [INDIRECT_PROMPT_INJECTION]: The skill processes codebase content and external documentation during its validation phases. While this exposes a surface for indirect prompt injection from potentially malicious code or external documentation, this is an inherent aspect of the skill's primary purpose (code validation). The risk is mitigated by the use of an isolated environment and a structured 6-phase workflow.- [COMMAND_EXECUTION]: The workflow includes running linters and codebase exploration tools. These activities are typical for development-oriented agents and are presented as integral components of the code-quality validation process.- [SAFE]: The agent specified (fuse-ai-pilot:sniper) aligns with the author's identity (fusengine), indicating the use of official vendor-provided resources.

sniper-check