SkillOpt v0.1.0: initial release

- Skill optimization framework with training loop analogy
- 11 benchmarks, 4 model backends (Azure OpenAI, Claude, Codex, Qwen)
- WebUI for browser-based training control
- Pluggable architecture for extending benchmarks and backends

skillopt/envs/sealqa/skills/initial.md

@@ -0,0 +1,11 @@
+# SealQA Skill
+
+## Evidence Gathering
+- Search for the most directly relevant evidence before answering.
+- If multiple sources conflict, prefer the source that best matches the question's entity, date, and scope.
+- Keep notes on which evidence directly answers the question versus which evidence is only contextual.
+
+## Final Answer Discipline
+- Do not answer until the supporting evidence is specific enough.
+- Choose the final answer that is best grounded in the gathered evidence.
+- Keep the final answer concise.