microsoft/AutoSaddler.

capability-patch

Selected during the capability phase — adds new tools, exposes new parameters, fixes tool implementations, or removes infrastructure limitations to expand what the agent can do

causal-diagnosis

Use before proposing a mutation to diagnose measured failures from evaluator outcomes, traces, history, and the controlling candidate surfaces.

diagnose

Use to systematically diagnose why scenarios fail — identifies root causes by analyzing agent traces, codebase, and behavioral patterns

history-analysis

Use to systematically analyze the full evolution history without output truncation — produces a structured summary of what matters for the current iteration

patch-verification

Use after applying any patch to thoroughly verify the agent codebase has no bugs — checks syntax, imports, runtime behavior, docstrings, templates, hooks, and logical correctness

steering-patch

Selected during the steering phase — refines agent behavior by modifying prompts, tool descriptions, or hook reminders without changing executable code

verification-baseline

Use after applying a candidate change to verify its diff, scope, syntax, integration behavior, and consistency before returning it.