agentv 0.17.0 → 0.18.0

package/dist/cli.js

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   runCli
-} from "./chunk-KL3Y2C6J.js";
+} from "./chunk-PHILDJ3W.js";
 import "./chunk-UE4GLFVL.js";
 
 // src/cli.ts

package/dist/cli.js.map

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

@@ -1,7 +1,7 @@
 import {
   createProgram,
   runCli
-} from "./chunk-KL3Y2C6J.js";
+} from "./chunk-PHILDJ3W.js";
 import "./chunk-UE4GLFVL.js";
 export {
   createProgram,

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.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "agentv",
-  "version": "0.17.0",
+  "version": "0.18.0",
   "description": "CLI entry point for AgentV",
   "type": "module",
   "repository": {
@@ -14,7 +14,10 @@
   "bin": {
     "agentv": "./dist/cli.js"
   },
-  "files": ["dist", "README.md"],
+  "files": [
+    "dist",
+    "README.md"
+  ],
   "scripts": {
     "dev": "bun --watch src/index.ts",
     "build": "tsup && bun run copy-readme",