The Agent Skills Directory

[SAFE]: The skill operates as a read-only research orchestrator, routing queries to specialized agents that use specific MCP tools (e.g., mcp_fusion_search_framework, mcp_fusion_search_eds). This architecture maintains a clear separation of concerns and follows the principle of least privilege by using targeted search indexes.
[SAFE]: Instructions throughout the skill emphasize returning only source-backed evidence and explicitly forbid the creation of code changes, installation of skills, or modification of repositories. Safety constraints are clearly defined in SKILL.md and reinforced in each sub-agent file (e.g., agents/backend-code.agent.md, agents/framework.agent.md).
[SAFE]: All external references, including those in the CHANGELOG.md, point to official Equinor GitHub repositories. No unauthorized or suspicious third-party domains were identified. The skill correctly identifies and uses vendor-owned resources.
[SAFE]: No obfuscation, prompt injection, or persistence mechanisms were found. The skill uses standardized templates for answers (assets/source-backed-answer-template.md) and follows a rigorous classification and dispatch process, reducing the risk of unintended behavior.

fusion-research