latch-eval-tools 0.1.1__tar.gz → 0.1.4__tar.gz

{latch_eval_tools-0.1.1 → latch_eval_tools-0.1.4}/PKG-INFO

@@ -1,11 +1,12 @@
 Metadata-Version: 2.4
 Name: latch-eval-tools
-Version: 0.1.1
+Version: 0.1.4
 Summary: Shared eval tools for single-cell bench, spatial bench, and future biology benchmarks.
 License-File: LICENSE
 Requires-Python: >=3.10
 Requires-Dist: aiohttp>=3.0.0
 Requires-Dist: anthropic>=0.72.0
+Requires-Dist: latch-config>=0.1.0
 Requires-Dist: latch>=2.0.0
 Requires-Dist: matplotlib>=3.0.0
 Requires-Dist: mini-swe-agent

{latch_eval_tools-0.1.1 → latch_eval_tools-0.1.4}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "latch-eval-tools"
-version = "0.1.1"
+version = "0.1.4"
 description = "Shared eval tools for single-cell bench, spatial bench, and future biology benchmarks."
 readme = "README.md"
 requires-python = ">=3.10"
@@ -21,6 +21,7 @@ dependencies = [
     "openai>=1.0.0",
     "anthropic>=0.72.0",
     "latch>=2.0.0",
+    "latch_config>=0.1.0",
     "mini-swe-agent",
 ]
 

{latch_eval_tools-0.1.1 → latch_eval_tools-0.1.4}/src/latch_eval_tools/harness/claudecode.py

@@ -38,9 +38,9 @@ def run_claudecode_task(
         agent_log_file.unlink()
 
     enhanced_prompt = _enhance_prompt_with_local_files(task_prompt, work_dir)
-    enhanced_prompt += """
+    enhanced_prompt += f"""
 
-CRITICAL: You must write eval_answer.json BEFORE signaling completion.
+CRITICAL: You must write eval_answer.json BEFORE signaling completion in the {work_dir}.
 Correct order: 1) Perform analysis 2) Write eval_answer.json with your answer 3) Exit"""
 
     cmd = ["claude", "--print", "--dangerously-skip-permissions", "--verbose", "--output-format", "stream-json"]
@@ -59,6 +59,7 @@ Correct order: 1) Perform analysis 2) Write eval_answer.json with your answer 3)
             home_dir = Path.home()
             current_mode = home_dir.stat().st_mode
             home_dir.chmod(current_mode | stat.S_IXOTH)
+
             eval_cache_dir = home_dir / ".eval_cache"
             if eval_cache_dir.exists():
                 shutil.chown(eval_cache_dir, user="claude", group="claude")
@@ -67,6 +68,8 @@ Correct order: 1) Perform analysis 2) Write eval_answer.json with your answer 3)
                         shutil.chown(item, user="claude", group="claude")
                     except PermissionError:
                         pass
+
+            work_dir.chmod(0o777)
         except KeyError:
             run_as_claude_user = False