Authoring role for the D365FO agent eval loop catalog. Drafts a new eval/cases/ .json spec (valid against eval/cases/schema.json), scaffolds its golden folder, and sets goldenpending until the golden is captured on the VM. Use when asked to "add an eval case", "author a case for ", "draft a case from this failure", or…
Implementer role of the D365FO agent eval loop. Runs ON THE VM (mcp-server in full mode + C# bridge) against the Contoso sandbox model. Takes an eval case id, drives the grounded MCP tool path to implement it, builds, scores against the golden/SysTest oracle, writes a corpus record, and rolls back. Use when asked to…
Improver role of the D365FO agent eval loop. Reads the corpus of run records, ranks failure clusters, reproduces a TOOLDEFECT/KNOWLEDGEGAP/VALIDATORGAP as a minimal VM-free repo test, fixes it, validates against the held-out split, and opens a PR citing corpus evidence. Runs in the repo (never touches the VM). Use…