agentv 0.16.0 → 0.18.0

package/dist/{chunk-7XYYGJAC.js → chunk-UE4GLFVL.js}

@@ -34,9 +34,9 @@ var __toESM = (mod, isNodeMode, target) => (target = mod != null ? __create(__ge
   mod
 ));
 
-// ../../node_modules/.pnpm/@vercel+oidc@3.0.5/node_modules/@vercel/oidc/dist/token-error.js
+// ../../node_modules/.bun/@vercel+oidc@3.0.5/node_modules/@vercel/oidc/dist/token-error.js
 var require_token_error = __commonJS({
-  "../../node_modules/.pnpm/@vercel+oidc@3.0.5/node_modules/@vercel/oidc/dist/token-error.js"(exports, module) {
+  "../../node_modules/.bun/@vercel+oidc@3.0.5/node_modules/@vercel/oidc/dist/token-error.js"(exports, module) {
     "use strict";
     var __defProp2 = Object.defineProperty;
     var __getOwnPropDesc2 = Object.getOwnPropertyDescriptor;
@@ -83,4 +83,4 @@ export {
   __toESM,
   require_token_error
 };
-//# sourceMappingURL=chunk-7XYYGJAC.js.map
+//# sourceMappingURL=chunk-UE4GLFVL.js.map

package/dist/chunk-UE4GLFVL.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../node_modules/.bun/@vercel+oidc@3.0.5/node_modules/@vercel/oidc/dist/token-error.js"],"sourcesContent":["\"use strict\";\nvar __defProp = Object.defineProperty;\nvar __getOwnPropDesc = Object.getOwnPropertyDescriptor;\nvar __getOwnPropNames = Object.getOwnPropertyNames;\nvar __hasOwnProp = Object.prototype.hasOwnProperty;\nvar __export = (target, all) => {\n  for (var name in all)\n    __defProp(target, name, { get: all[name], enumerable: true });\n};\nvar __copyProps = (to, from, except, desc) => {\n  if (from && typeof from === \"object\" || typeof from === \"function\") {\n    for (let key of __getOwnPropNames(from))\n      if (!__hasOwnProp.call(to, key) && key !== except)\n        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });\n  }\n  return to;\n};\nvar __toCommonJS = (mod) => __copyProps(__defProp({}, \"__esModule\", { value: true }), mod);\nvar token_error_exports = {};\n__export(token_error_exports, {\n  VercelOidcTokenError: () => VercelOidcTokenError\n});\nmodule.exports = __toCommonJS(token_error_exports);\nclass VercelOidcTokenError extends Error {\n  constructor(message, cause) {\n    super(message);\n    this.name = \"VercelOidcTokenError\";\n    this.cause = cause;\n  }\n  toString() {\n    if (this.cause) {\n      return `${this.name}: ${this.message}: ${this.cause}`;\n    }\n    return `${this.name}: ${this.message}`;\n  }\n}\n// Annotate the CommonJS export names for ESM import in node:\n0 && (module.exports = {\n  VercelOidcTokenError\n});\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AACA,QAAIA,aAAY,OAAO;AACvB,QAAIC,oBAAmB,OAAO;AAC9B,QAAIC,qBAAoB,OAAO;AAC/B,QAAIC,gBAAe,OAAO,UAAU;AACpC,QAAIC,YAAW,CAAC,QAAQ,QAAQ;AAC9B,eAAS,QAAQ;AACf,QAAAJ,WAAU,QAAQ,MAAM,EAAE,KAAK,IAAI,IAAI,GAAG,YAAY,KAAK,CAAC;AAAA,IAChE;AACA,QAAIK,eAAc,CAAC,IAAI,MAAM,QAAQ,SAAS;AAC5C,UAAI,QAAQ,OAAO,SAAS,YAAY,OAAO,SAAS,YAAY;AAClE,iBAAS,OAAOH,mBAAkB,IAAI;AACpC,cAAI,CAACC,cAAa,KAAK,IAAI,GAAG,KAAK,QAAQ;AACzC,YAAAH,WAAU,IAAI,KAAK,EAAE,KAAK,MAAM,KAAK,GAAG,GAAG,YAAY,EAAE,OAAOC,kBAAiB,MAAM,GAAG,MAAM,KAAK,WAAW,CAAC;AAAA,MACvH;AACA,aAAO;AAAA,IACT;AACA,QAAI,eAAe,CAAC,QAAQI,aAAYL,WAAU,CAAC,GAAG,cAAc,EAAE,OAAO,KAAK,CAAC,GAAG,GAAG;AACzF,QAAI,sBAAsB,CAAC;AAC3B,IAAAI,UAAS,qBAAqB;AAAA,MAC5B,sBAAsB,MAAM;AAAA,IAC9B,CAAC;AACD,WAAO,UAAU,aAAa,mBAAmB;AACjD,QAAM,uBAAN,cAAmC,MAAM;AAAA,MACvC,YAAY,SAAS,OAAO;AAC1B,cAAM,OAAO;AACb,aAAK,OAAO;AACZ,aAAK,QAAQ;AAAA,MACf;AAAA,MACA,WAAW;AACT,YAAI,KAAK,OAAO;AACd,iBAAO,GAAG,KAAK,IAAI,KAAK,KAAK,OAAO,KAAK,KAAK,KAAK;AAAA,QACrD;AACA,eAAO,GAAG,KAAK,IAAI,KAAK,KAAK,OAAO;AAAA,MACtC;AAAA,IACF;AAAA;AAAA;","names":["__defProp","__getOwnPropDesc","__getOwnPropNames","__hasOwnProp","__export","__copyProps"]}

package/dist/cli.js

@@ -1,8 +1,8 @@
 #!/usr/bin/env node
 import {
   runCli
-} from "./chunk-HWGALLUR.js";
-import "./chunk-7XYYGJAC.js";
+} from "./chunk-PHILDJ3W.js";
+import "./chunk-UE4GLFVL.js";
 
 // src/cli.ts
 void runCli();

package/dist/cli.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/cli.ts"],"sourcesContent":["#!/usr/bin/env node\r\nimport { runCli } from './index.js';\r\n\r\nvoid runCli();\r\n"],"mappings":";;;;;;;AAGA,KAAK,OAAO;","names":[]}
+{"version":3,"sources":["../src/cli.ts"],"sourcesContent":["#!/usr/bin/env node\r\nimport { runCli } from \"./index.js\";\r\n\r\nvoid runCli();\r\n"],"mappings":";;;;;;;AAGA,KAAK,OAAO;","names":[]}

package/dist/index.js

@@ -1,8 +1,8 @@
 import {
   createProgram,
   runCli
-} from "./chunk-HWGALLUR.js";
-import "./chunk-7XYYGJAC.js";
+} from "./chunk-PHILDJ3W.js";
+import "./chunk-UE4GLFVL.js";
 export {
   createProgram,
   runCli

package/dist/templates/.claude/skills/agentv-eval-builder/references/config-schema.json

@@ -1,27 +1,27 @@
-{
-  "$schema": "http://json-schema.org/draft-07/schema#",
-  "title": "AgentV Config Schema",
-  "description": "Schema for .agentv/config.yaml configuration files",
-  "type": "object",
-  "properties": {
-    "$schema": {
-      "type": "string",
-      "description": "Schema identifier",
-      "enum": ["agentv-config-v2"]
-    },
-    "guideline_patterns": {
-      "type": "array",
-      "description": "Glob patterns for identifying guideline files (instructions, prompts). Files matching these patterns are treated as guidelines, while non-matching files are treated as regular file content.",
-      "items": {
-        "type": "string",
-        "description": "Glob pattern (e.g., '**/*.instructions.md', '**/prompts/**')"
-      },
-      "examples": [
-        ["**/*.instructions.md", "**/instructions/**", "**/*.prompt.md", "**/prompts/**"],
-        ["**/*.guide.md", "**/guidelines/**", "docs/AGENTS.md"]
-      ]
-    }
-  },
-  "required": ["$schema"],
-  "additionalProperties": false
-}
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "title": "AgentV Config Schema",
+  "description": "Schema for .agentv/config.yaml configuration files",
+  "type": "object",
+  "properties": {
+    "$schema": {
+      "type": "string",
+      "description": "Schema identifier",
+      "enum": ["agentv-config-v2"]
+    },
+    "guideline_patterns": {
+      "type": "array",
+      "description": "Glob patterns for identifying guideline files (instructions, prompts). Files matching these patterns are treated as guidelines, while non-matching files are treated as regular file content.",
+      "items": {
+        "type": "string",
+        "description": "Glob pattern (e.g., '**/*.instructions.md', '**/prompts/**')"
+      },
+      "examples": [
+        ["**/*.instructions.md", "**/instructions/**", "**/*.prompt.md", "**/prompts/**"],
+        ["**/*.guide.md", "**/guidelines/**", "docs/AGENTS.md"]
+      ]
+    }
+  },
+  "required": ["$schema"],
+  "additionalProperties": false
+}

package/dist/templates/.claude/skills/agentv-eval-builder/references/eval-schema.json

@@ -1,217 +1,217 @@
-{
-  "$schema": "http://json-schema.org/draft-07/schema#",
-  "title": "AgentV Eval Schema",
-  "description": "Schema for YAML evaluation files with conversation flows, multiple evaluators, and execution configuration",
-  "type": "object",
-  "properties": {
-    "$schema": {
-      "type": "string",
-      "description": "Schema identifier",
-      "enum": ["agentv-eval-v2"]
-    },
-    "description": {
-      "type": "string",
-      "description": "Description of what this eval suite covers"
-    },
-    "target": {
-      "type": "string",
-      "description": "(Deprecated: use execution.target instead) Default target configuration name. Can be overridden per eval case."
-    },
-    "execution": {
-      "type": "object",
-      "description": "Default execution configuration for all eval cases (can be overridden per case)",
-      "properties": {
-        "target": {
-          "type": "string",
-          "description": "Default target configuration name (e.g., default, azure_base, vscode_projectx). Can be overridden per eval case."
-        },
-        "evaluators": {
-          "type": "array",
-          "description": "Default evaluators for all eval cases (code-based and LLM judges)",
-          "items": {
-            "type": "object",
-            "properties": {
-              "name": {
-                "type": "string",
-                "description": "Evaluator name/identifier"
-              },
-              "type": {
-                "type": "string",
-                "enum": ["code", "llm_judge"],
-                "description": "Evaluator type: 'code' for scripts/regex/keywords, 'llm_judge' for LLM-based evaluation"
-              },
-              "script": {
-                "type": "string",
-                "description": "Path to evaluator script (for type: code)"
-              },
-              "prompt": {
-                "type": "string",
-                "description": "Path to judge prompt file (for type: llm_judge)"
-              }
-            },
-            "required": ["name", "type"],
-            "additionalProperties": true
-          }
-        }
-      },
-      "additionalProperties": true
-    },
-    "evalcases": {
-      "type": "array",
-      "description": "Array of evaluation cases",
-      "minItems": 1,
-      "items": {
-        "type": "object",
-        "properties": {
-          "id": {
-            "type": "string",
-            "description": "Unique identifier for the eval case"
-          },
-          "conversation_id": {
-            "type": "string",
-            "description": "Optional conversation identifier for threading multiple eval cases together"
-          },
-          "outcome": {
-            "type": "string",
-            "description": "Description of what the AI should accomplish in this eval"
-          },
-          "note": {
-            "type": "string",
-            "description": "Optional note or additional context for the eval case. Use this to document test-specific considerations, known limitations, or rationale for expected behavior."
-          },
-          "input_messages": {
-            "type": "array",
-            "description": "Input messages for the conversation",
-            "minItems": 1,
-            "items": {
-              "type": "object",
-              "properties": {
-                "role": {
-                  "type": "string",
-                  "enum": ["system", "user", "assistant", "tool"],
-                  "description": "Message role"
-                },
-                "content": {
-                  "oneOf": [
-                    {
-                      "type": "string",
-                      "description": "Simple text content"
-                    },
-                    {
-                      "type": "array",
-                      "description": "Mixed content items (text and file references)",
-                      "items": {
-                        "type": "object",
-                        "properties": {
-                          "type": {
-                            "type": "string",
-                            "enum": ["text", "file"],
-                            "description": "Content type: 'text' for inline content, 'file' for file references"
-                          },
-                          "value": {
-                            "type": "string",
-                            "description": "Text content or file path. Relative paths (e.g., ../prompts/file.md) are resolved from eval file directory. Absolute paths (e.g., /docs/examples/prompts/file.md) are resolved from repo root."
-                          }
-                        },
-                        "required": ["type", "value"],
-                        "additionalProperties": false
-                      }
-                    }
-                  ]
-                }
-              },
-              "required": ["role", "content"],
-              "additionalProperties": false
-            }
-          },
-          "expected_messages": {
-            "type": "array",
-            "description": "Expected response messages",
-            "minItems": 1,
-            "items": {
-              "type": "object",
-              "properties": {
-                "role": {
-                  "type": "string",
-                  "enum": ["system", "user", "assistant", "tool"],
-                  "description": "Message role"
-                },
-                "content": {
-                  "oneOf": [
-                    {
-                      "type": "string",
-                      "description": "Simple text content"
-                    },
-                    {
-                      "type": "array",
-                      "description": "Mixed content items",
-                      "items": {
-                        "type": "object",
-                        "properties": {
-                          "type": {
-                            "type": "string",
-                            "enum": ["text", "file"]
-                          },
-                          "value": {
-                            "type": "string"
-                          }
-                        },
-                        "required": ["type", "value"],
-                        "additionalProperties": false
-                      }
-                    }
-                  ]
-                }
-              },
-              "required": ["role", "content"],
-              "additionalProperties": false
-            }
-          },
-          "execution": {
-            "type": "object",
-            "description": "Per-case execution configuration",
-            "properties": {
-              "target": {
-                "type": "string",
-                "description": "Override target for this specific eval case"
-              },
-              "evaluators": {
-                "type": "array",
-                "description": "Multiple evaluators (code-based and LLM judges)",
-                "items": {
-                  "type": "object",
-                  "properties": {
-                    "name": {
-                      "type": "string",
-                      "description": "Evaluator name/identifier"
-                    },
-                    "type": {
-                      "type": "string",
-                      "enum": ["code", "llm_judge"],
-                      "description": "Evaluator type: 'code' for scripts/regex/keywords, 'llm_judge' for LLM-based evaluation"
-                    },
-                    "script": {
-                      "type": "string",
-                      "description": "Path to evaluator script (for type: code)"
-                    },
-                    "prompt": {
-                      "type": "string",
-                      "description": "Path to judge prompt file (for type: llm_judge)"
-                    }
-                  },
-                  "required": ["name", "type"],
-                  "additionalProperties": true
-                }
-              }
-            },
-            "additionalProperties": true
-          }
-        },
-        "required": ["id", "outcome", "input_messages", "expected_messages"],
-        "additionalProperties": false
-      }
-    }
-  },
-  "required": ["evalcases"],
-  "additionalProperties": false
-}
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "title": "AgentV Eval Schema",
+  "description": "Schema for YAML evaluation files with conversation flows, multiple evaluators, and execution configuration",
+  "type": "object",
+  "properties": {
+    "$schema": {
+      "type": "string",
+      "description": "Schema identifier",
+      "enum": ["agentv-eval-v2"]
+    },
+    "description": {
+      "type": "string",
+      "description": "Description of what this eval suite covers"
+    },
+    "target": {
+      "type": "string",
+      "description": "(Deprecated: use execution.target instead) Default target configuration name. Can be overridden per eval case."
+    },
+    "execution": {
+      "type": "object",
+      "description": "Default execution configuration for all eval cases (can be overridden per case)",
+      "properties": {
+        "target": {
+          "type": "string",
+          "description": "Default target configuration name (e.g., default, azure_base, vscode_projectx). Can be overridden per eval case."
+        },
+        "evaluators": {
+          "type": "array",
+          "description": "Default evaluators for all eval cases (code-based and LLM judges)",
+          "items": {
+            "type": "object",
+            "properties": {
+              "name": {
+                "type": "string",
+                "description": "Evaluator name/identifier"
+              },
+              "type": {
+                "type": "string",
+                "enum": ["code", "llm_judge"],
+                "description": "Evaluator type: 'code' for scripts/regex/keywords, 'llm_judge' for LLM-based evaluation"
+              },
+              "script": {
+                "type": "string",
+                "description": "Path to evaluator script (for type: code)"
+              },
+              "prompt": {
+                "type": "string",
+                "description": "Path to judge prompt file (for type: llm_judge)"
+              }
+            },
+            "required": ["name", "type"],
+            "additionalProperties": true
+          }
+        }
+      },
+      "additionalProperties": true
+    },
+    "evalcases": {
+      "type": "array",
+      "description": "Array of evaluation cases",
+      "minItems": 1,
+      "items": {
+        "type": "object",
+        "properties": {
+          "id": {
+            "type": "string",
+            "description": "Unique identifier for the eval case"
+          },
+          "conversation_id": {
+            "type": "string",
+            "description": "Optional conversation identifier for threading multiple eval cases together"
+          },
+          "outcome": {
+            "type": "string",
+            "description": "Description of what the AI should accomplish in this eval"
+          },
+          "note": {
+            "type": "string",
+            "description": "Optional note or additional context for the eval case. Use this to document test-specific considerations, known limitations, or rationale for expected behavior."
+          },
+          "input_messages": {
+            "type": "array",
+            "description": "Input messages for the conversation",
+            "minItems": 1,
+            "items": {
+              "type": "object",
+              "properties": {
+                "role": {
+                  "type": "string",
+                  "enum": ["system", "user", "assistant", "tool"],
+                  "description": "Message role"
+                },
+                "content": {
+                  "oneOf": [
+                    {
+                      "type": "string",
+                      "description": "Simple text content"
+                    },
+                    {
+                      "type": "array",
+                      "description": "Mixed content items (text and file references)",
+                      "items": {
+                        "type": "object",
+                        "properties": {
+                          "type": {
+                            "type": "string",
+                            "enum": ["text", "file"],
+                            "description": "Content type: 'text' for inline content, 'file' for file references"
+                          },
+                          "value": {
+                            "type": "string",
+                            "description": "Text content or file path. Relative paths (e.g., ../prompts/file.md) are resolved from eval file directory. Absolute paths (e.g., /docs/examples/prompts/file.md) are resolved from repo root."
+                          }
+                        },
+                        "required": ["type", "value"],
+                        "additionalProperties": false
+                      }
+                    }
+                  ]
+                }
+              },
+              "required": ["role", "content"],
+              "additionalProperties": false
+            }
+          },
+          "expected_messages": {
+            "type": "array",
+            "description": "Expected response messages",
+            "minItems": 1,
+            "items": {
+              "type": "object",
+              "properties": {
+                "role": {
+                  "type": "string",
+                  "enum": ["system", "user", "assistant", "tool"],
+                  "description": "Message role"
+                },
+                "content": {
+                  "oneOf": [
+                    {
+                      "type": "string",
+                      "description": "Simple text content"
+                    },
+                    {
+                      "type": "array",
+                      "description": "Mixed content items",
+                      "items": {
+                        "type": "object",
+                        "properties": {
+                          "type": {
+                            "type": "string",
+                            "enum": ["text", "file"]
+                          },
+                          "value": {
+                            "type": "string"
+                          }
+                        },
+                        "required": ["type", "value"],
+                        "additionalProperties": false
+                      }
+                    }
+                  ]
+                }
+              },
+              "required": ["role", "content"],
+              "additionalProperties": false
+            }
+          },
+          "execution": {
+            "type": "object",
+            "description": "Per-case execution configuration",
+            "properties": {
+              "target": {
+                "type": "string",
+                "description": "Override target for this specific eval case"
+              },
+              "evaluators": {
+                "type": "array",
+                "description": "Multiple evaluators (code-based and LLM judges)",
+                "items": {
+                  "type": "object",
+                  "properties": {
+                    "name": {
+                      "type": "string",
+                      "description": "Evaluator name/identifier"
+                    },
+                    "type": {
+                      "type": "string",
+                      "enum": ["code", "llm_judge"],
+                      "description": "Evaluator type: 'code' for scripts/regex/keywords, 'llm_judge' for LLM-based evaluation"
+                    },
+                    "script": {
+                      "type": "string",
+                      "description": "Path to evaluator script (for type: code)"
+                    },
+                    "prompt": {
+                      "type": "string",
+                      "description": "Path to judge prompt file (for type: llm_judge)"
+                    }
+                  },
+                  "required": ["name", "type"],
+                  "additionalProperties": true
+                }
+              }
+            },
+            "additionalProperties": true
+          }
+        },
+        "required": ["id", "outcome", "input_messages", "expected_messages"],
+        "additionalProperties": false
+      }
+    }
+  },
+  "required": ["evalcases"],
+  "additionalProperties": false
+}

package/dist/templates/.claude/skills/agentv-prompt-optimizer/SKILL.md

@@ -21,41 +21,48 @@ description: Iteratively optimize prompt files against AgentV evaluation dataset
         - If not, create a new one in the parent directory of the eval files: `optimization-[timestamp].md`.
     - Read content of the identified prompt file.
 
-2.  **Optimization Loop** (Max 5 iterations)
-    - **Execute (The Generator)**: Run `pnpm agentv eval <eval-path>`.
+2.  **Optimization Loop** (Max 10 iterations)
+    - **Execute (The Generator)**: Run `agentv eval <eval-path>`.
+        - *Targeted Run*: If iterating on specific stubborn failures, use `--eval-id <case_id>` to run only the relevant eval cases.
     - **Analyze (The Reflector)**:
         - Locate the results file path from the console output (e.g., `.agentv/results/eval_...jsonl`).
-        - Read the results file. Calculate pass rate.
-        - **Root Cause Analysis**: For each failure, perform a deep dive:
-            - **Error Identification**: What exactly went wrong? (e.g., "Predicted 'High' but expected 'Low'")
-            - **Root Cause**: Why did it happen? (e.g., "Ambiguous definition of 'High' severity", "Hallucinated a constraint", "Incorrect test expectation")
-            - **Correct Approach**: What *should* the model have done?
-            - **Key Insight**: What general rule or pattern can we learn from this?
-            - **Regression Check**: Did this change break previously passing tests? If so, mark the previous change as "Harmful".
+        - **Orchestrate Subagent**: Use `runSubagent` to analyze the results.
+            - **Task**: Read the results file, calculate pass rate, and perform root cause analysis.
+            - **Output**: Return a structured analysis including:
+                - **Score**: Current pass rate.
+                - **Root Cause**: Why failures occurred (e.g., "Ambiguous definition", "Hallucination").
+                - **Insight**: Key learning or pattern identified from the failures.
+                - **Strategy**: High-level plan to fix the prompt (e.g., "Clarify section X", "Add negative constraint").
     - **Decide**:
         - If **100% pass**: STOP and report success.
         - If **Score decreased**: Revert last change, try different approach.
         - If **No improvement** (2x): STOP and report stagnation.
-    - **Log Result**:
-        - Append the result of this iteration to the identified optimization log file.
-        - **Format**:
-          ```markdown
-          ### Iteration [N]
-          - **Change**: [Description of edit]
-          - **Rationale**: [Root Cause / Why this fix was chosen]
-          - **Outcome**: [Success / Failure / Harmful] (Score: X% -> Y%)
-          - **Insight**: [Key learning or pattern identified]
-          ```
     - **Refine (The Curator)**:
-        - Modify the relevant `<prompt-file>` (pick the one most likely to be the root cause) to address failures.
-        - **Strategy**: Treat the prompt as a structured set of rules and instructions.
-            - **Clarify**: If ambiguous, make the existing instruction more specific.
-            - **Add Rule**: If a constraint was missed, add a specific bullet point to the relevant section.
-            - **Negative Constraint**: If hallucinating, explicitly state what NOT to do.
-            - **Consolidate**: Check for redundant or overlapping instructions and merge them.
+        - **Orchestrate Subagent**: Use `runSubagent` to apply the fix.
+            - **Task**: Read the relevant prompt file(s), apply the **Strategy** from the Reflector, and generate the log entry.
+            - **Output**: The **Log Entry** describing the specific operation performed.
+                  ```markdown
+                  ### Iteration [N]
+                  - **Operation**: [ADD / UPDATE / DELETE]
+                  - **Target**: [Section Name]
+                  - **Change**: [Specific text added/modified]
+                  - **Trigger**: [Specific failing test case or error pattern]
+                  - **Rationale**: [From Reflector: Root Cause]
+                  - **Score**: [From Reflector: Current Pass Rate]
+                  - **Insight**: [From Reflector: Key Learning]
+                  ```
+        - **Strategy**: Treat the prompt as a structured set of rules. Execute atomic operations:
+            - **ADD**: Insert a new rule if a constraint was missed.
+            - **UPDATE**: Refine an existing rule to be clearer or more general.
+                - *Clarify*: Make ambiguous instructions specific.
+                - *Generalize*: Refactor specific fixes into high-level principles (First Principles).
+            - **DELETE**: Remove obsolete, redundant, or harmful rules.
+                - *Prune*: If a general rule covers specific cases, delete the specific ones.
+            - **Negative Constraint**: If hallucinating, explicitly state what NOT to do. Prefer generalized prohibitions over specific forbidden tokens where possible.
             - **Safety Check**: Ensure new rules don't contradict existing ones (unless intended).
         - **Constraint**: Avoid rewriting large sections. Make surgical, additive changes to preserve existing behavior.
-        - **Apply**: Use `replace_string_in_file`.
+    - **Log Result**:
+        - Append the **Log Entry** returned by the Curator to the optimization log file.
 
 3.  **Completion**
     - Report final score.
@@ -63,7 +70,8 @@ description: Iteratively optimize prompt files against AgentV evaluation dataset
     - **Finalize Optimization Log**: Add a summary header to the optimization log file indicating the session completion and final score.
 
 ## Guidelines
-- **Simplicity ("Less is More")**: Avoid adding specific rules for rare edge cases ("hotfixes"). Focus on universally applicable instructions.
+- **Generalization First**: Prefer broad, principle-based guidelines over specific examples or "hotfixes". Only use specific rules if generalized instructions fail to achieve the desired score.
+- **Simplicity ("Less is More")**: Avoid overfitting to the test set. If a specific rule doesn't significantly improve the score compared to a general one, choose the general one.
 - **Structure**: Maintain existing Markdown headers/sections.
 - **Progressive Disclosure**: If the prompt grows too large (>200 lines), consider moving specialized logic into a separate file or skill.
 - **Quality Criteria**: Ensure the prompt defines a clear persona, specific task, and measurable success criteria.