Remove dead code and deduplicate agent prompt resolution

core/decision.py

@@ -153,11 +153,6 @@ def execute_decision_step(step: DecisionStep, step_num: int) -> tuple[int, str,
     """
     print(f"Executing decision step: {step.name}")
 
-    # # Check that we have at least one decision
-    # if not step.decision:
-    #     print("⚠ No decisions defined, defaulting to CONTINUE")
-    #     return (SUCCESS, "CONTINUE", None)
-
     # Get the mode from the first decision (all decisions in a step should have the same mode)
     decision_mode = step.mode
     print(f"Decision mode: {decision_mode}")

core/task.py

@@ -776,6 +776,26 @@ def replace_field(match):
     return Path(path)
 
 
+def _resolve_agent_prompt(prompt: str, framework_root: str) -> str:
+    """Resolve agent file references in prompts, loading and injecting agent content."""
+    agent_file_match = re.search(r'\./(?:\.opencode|prompts)/(agents?)/([^.\s]+)\.md', prompt)
+    if not agent_file_match:
+        return prompt
+
+    prompts_dir = os.environ.get('PROMPTS_DIR', str(Path(framework_root) / 'prompts'))
+    agent_file_path = Path(prompts_dir) / agent_file_match.group(1) / f"{agent_file_match.group(2)}.md"
+    if not agent_file_path.exists():
+        print(f"  ⚠ Agent file not found: {agent_file_path}")
+        return prompt
+
+    print(f"  Loading agent definition from: {agent_file_path}")
+    agent_content = agent_file_path.read_text()
+    parts = agent_content.split('---', 2)
+    if len(parts) >= 3:
+        return parts[2].strip()
+    return agent_content
+
+
 def execute_task_step(step: TaskStep, step_num: int, step_id: str | None = None) -> tuple[int, str, str | None, str | None]:
     """
     Execute a task step based on its model type.
@@ -913,26 +933,7 @@ def execute_task_step(step: TaskStep, step_num: int, step_id: str | None = None)
         parser_script_file = Path(framework_root) / 'log_formatters' / 'claude_code.py'
 
         # Process prompt - if it references an agent file, read and inject its content
-        prompt = step.prompt
-
-        agent_file_match = re.search(r'\./(?:\.opencode|prompts)/(agents?)/([^.\s]+)\.md', prompt)
-        if agent_file_match:
-            # Resolve agent file from PROMPTS_DIR
-            prompts_dir = os.environ.get('PROMPTS_DIR', str(Path(framework_root) / 'prompts'))
-            agent_file_path = Path(prompts_dir) / agent_file_match.group(1) / f"{agent_file_match.group(2)}.md"
-            if agent_file_path.exists():
-                print(f"  Loading agent definition from: {agent_file_path}")
-                agent_content = agent_file_path.read_text()
-                # Extract content after frontmatter (between --- markers)
-                parts = agent_content.split('---', 2)
-                if len(parts) >= 3:
-                    # Use everything after the second ---
-                    prompt = parts[2].strip()
-                else:
-                    # No frontmatter, use full content
-                    prompt = agent_content
-            else:
-                print(f"  ⚠ Agent file not found: {agent_file_path}")
+        prompt = _resolve_agent_prompt(step.prompt, framework_root)
 
         # If repo_path is set, prefix command to cd there first
         # Use RECON_FOUNDRY_ROOT (where foundry.toml lives) with fallback to RECON_REPO_PATH
@@ -991,25 +992,7 @@ def execute_task_step(step: TaskStep, step_num: int, step_id: str | None = None)
         parser_script_file = Path(framework_root) / 'log_formatters' / 'opencode.py'
 
         # Process prompt - if it references an agent file, read and inject its content
-        prompt = step.prompt
-        agent_file_match = re.search(r'\./(?:\.opencode|prompts)/(agents?)/([^.\s]+)\.md', prompt)
-        if agent_file_match:
-            # Resolve agent file from PROMPTS_DIR
-            prompts_dir = os.environ.get('PROMPTS_DIR', str(Path(framework_root) / 'prompts'))
-            agent_file_path = Path(prompts_dir) / agent_file_match.group(1) / f"{agent_file_match.group(2)}.md"
-            if agent_file_path.exists():
-                print(f"  Loading agent definition from: {agent_file_path}")
-                agent_content = agent_file_path.read_text()
-                # Extract content after frontmatter (between --- markers)
-                parts = agent_content.split('---', 2)
-                if len(parts) >= 3:
-                    # Use everything after the second ---
-                    prompt = parts[2].strip()
-                else:
-                    # No frontmatter, use full content
-                    prompt = agent_content
-            else:
-                print(f"  ⚠ Agent file not found: {agent_file_path}")
+        prompt = _resolve_agent_prompt(step.prompt, framework_root)
 
         # If repo_path is set, prefix command to cd there first
         # Use RECON_FOUNDRY_ROOT (where foundry.toml lives) with fallback to RECON_REPO_PATH
@@ -1143,7 +1126,7 @@ def execute_task_step(step: TaskStep, step_num: int, step_id: str | None = None)
             error_body = ""
             try:
                 error_body = e.response.json().get("message", str(e))
-            except:
+            except Exception:
                 error_body = str(e)
             print(f"  ❌ Failed to dispatch fuzzing job: {error_body}")
             return (FAILURE, "CONTINUE", None, None)

server/utils.py

@@ -1,22 +1,7 @@
 """
-Utility functions for logging, parsing, and dependency checking.
+Utility functions for parsing.
 """
 
-import subprocess
-import time
-
-
-def create_log_file() -> str:
-    """
-    Create a timestamped log file path.
-    - Generate log ID from current timestamp
-    - Create path: artifacts/claude_logs_{timestamp}.json
-    - Return log file path
-    """
-    log_id = time.strftime("%Y%m%d_%H%M%S")
-    log_file = f"artifacts/claude_logs_{log_id}.json"
-    return log_file
-
 
 def parse_repo_info(repo_url: str) -> tuple[str, str]:
     """
@@ -40,39 +25,3 @@ def parse_repo_info(repo_url: str) -> tuple[str, str]:
     except Exception as e:
         print(f"Error parsing repo info: {e}")
         return "", ""
-
-
-def check_dependencies() -> dict[str, bool]:
-    """
-    Check if required tools are installed.
-    - Check: forge, echidna, medusa, halmos, node, claude
-    - Return dict of {tool: is_installed}
-    """
-    tools = ["forge", "echidna", "medusa", "halmos", "node", "claude", "git"]
-    status = {}
-
-    for tool in tools:
-        try:
-            result = subprocess.run(
-                ["which", tool],
-                capture_output=True,
-                text=True
-            )
-            status[tool] = result.returncode == 0
-        except Exception:
-            status[tool] = False
-
-    return status
-
-
-def install_missing_dependencies(deps: list[str]) -> bool:
-    """
-    Install missing dependencies.
-    - Call appropriate installers for each missing dependency
-    - Return overall success status
-    """
-    # This is a placeholder - actual implementation would need
-    # to handle different installation methods per dependency
-    print(f"Missing dependencies: {deps}")
-    print("Please install them manually or run the install_deps.sh script")
-    return False