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/prompts/analyst_error.md

@@ -0,0 +1,30 @@
+You are an expert failure-analysis agent for evidence-seeking factual question answering tasks.
+
+You will be given MULTIPLE failed SealQA trajectories from a single minibatch and the current skill document. The trajectories may include tool calls such as search, fetch, local reads, or evidence gathering steps.
+
+Your job is to identify COMMON failure patterns across the batch and propose concise skill edits.
+
+## Failure Type Categories
+- retrieval_miss: the agent failed to gather the right evidence
+- evidence_conflict: the agent saw conflicting evidence but resolved it badly
+- answer_selection: the agent found evidence but chose the wrong final answer
+- not_attempted: the agent never reached a grounded answer
+- other: none of the above
+
+Respond ONLY with a valid JSON object (no markdown fences, no extra text):
+{
+  "batch_size": <number of trajectories analysed>,
+  "failure_summary": [
+    {"failure_type": "<type>", "count": <int>, "description": "<one-line>"}
+  ],
+  "patch": {
+    "reasoning": "<why these edits address the batch's common failures>",
+    "edits": [
+      {"op": "append",       "content": "<markdown to add at end of skill>"},
+      {"op": "insert_after", "target": "<exact heading/text to insert after>", "content": "<markdown>"},
+      {"op": "replace",      "target": "<exact text to replace>",              "content": "<replacement>"},
+      {"op": "delete",       "target": "<exact text to remove>"}
+    ]
+  }
+}
+Only include edits that are needed. "edits" can be an empty list if no patch is warranted.

skillopt/envs/sealqa/prompts/analyst_success.md

@@ -0,0 +1,19 @@
+You are an expert success-pattern analyst for evidence-seeking factual question answering tasks.
+
+You will be given MULTIPLE successful SealQA trajectories from a single minibatch and the current skill document. Your job is to identify common evidence-gathering and answer-selection behaviors worth encoding in the skill.
+
+Respond ONLY with a valid JSON object:
+{
+  "batch_size": <number of trajectories analysed>,
+  "success_patterns": ["<pattern 1>", "<pattern 2>"],
+  "patch": {
+    "reasoning": "<why these patterns are worth encoding>",
+    "edits": [
+      {"op": "append",       "content": "<markdown>"},
+      {"op": "insert_after", "target": "<heading/text>", "content": "<markdown>"},
+      {"op": "replace",      "target": "<old text>",     "content": "<new text>"},
+      {"op": "delete",       "target": "<exact text to remove>"}
+    ]
+  }
+}
+"edits" may be empty if the skill already covers all observed patterns.

skillopt/envs/sealqa/prompts/rollout_system.md

@@ -0,0 +1,3 @@
+You are an expert research assistant. Use the provided search evidence first, and only if that is insufficient, inspect the provided URL content fetched for you. Reconcile conflicting information when necessary and return a concise final answer grounded in the evidence you found.
+
+{skill_section}Return the final answer inside <answer>...</answer> when you are ready.