Archive BUG-000022 evaluation and deferred final-review findings

tests/fixtures/llm_rejections/20261025/bug22-source-context-prefix/analysis.md

@@ -0,0 +1,5 @@
+# BUG-000022: missing source.context path_prefix
+
+The startup model requested the contextual source tool for Jankie using text 36410 and entity_refs N0002. The query omitted required path_prefix and was rejected with INVESTIGATION_OUTPUT. The next model call used plain source.search, producing mostly historical SQL and lexical numeric matches, with skipped files and partial coverage.
+
+Against the current protocol-2 query schema, the sole error is missing path_prefix. Adding the explicit empty prefix passes this query-schema check. This is a counterprobe, not an implemented automatic repair or proof of the full response's semantic validity. A future bounded syntax/default recovery could preserve the selected richer tool. Do not infer that lexical numeric search results exclude all dynamic callers.

tests/fixtures/llm_rejections/20261025/bug22-source-context-prefix/parsed-answer.json

@@ -0,0 +1,15 @@
+{
+  "assessment": null,
+  "next": {
+    "arguments": {
+      "entity_refs": [
+        "N0002"
+      ],
+      "kind": "source.context",
+      "text": "36410"
+    },
+    "kind": "tool",
+    "operation": "project.query"
+  },
+  "observation": "Search codebase context for references to creature entry 36410 (Officer Jankie) to evaluate potential scripted spawns, movement, or event handling."
+}

tests/fixtures/llm_rejections/20261025/bug22-source-context-prefix/query-schema.json

@@ -0,0 +1,39 @@
+{
+  "type": "object",
+  "additionalProperties": false,
+  "required": [
+    "kind",
+    "text",
+    "path_prefix"
+  ],
+  "properties": {
+    "kind": {
+      "const": "source.context"
+    },
+    "text": {
+      "type": "string",
+      "minLength": 1,
+      "maxLength": 200
+    },
+    "path_prefix": {
+      "type": "string",
+      "maxLength": 250
+    },
+    "scope": {
+      "enum": [
+        "code",
+        "schema",
+        "historical_sql"
+      ]
+    },
+    "entity_refs": {
+      "type": "array",
+      "maxItems": 4,
+      "uniqueItems": true,
+      "items": {
+        "type": "string",
+        "pattern": "^N[0-9]{4,}$"
+      }
+    }
+  }
+}