fix: address qodo review feedback

- remind_skill_continuation.py: use tempfile.gettempdir() for debug log
  path (cross-platform) and sys.stdout.write instead of print (T201).
- require_delegation.py: exit 1 on unexpected exception per hook
  exit-code policy (still non-blocking for the tool call itself).
- validate_task_graph_compliance.py, validate_task_graph_depth.py: add
  PEP 723 script metadata block (required for uv run --script).
- Delete orphaned system-prompts/workflow_orchestrator.md. No Python
  code loads this file; the full orchestrator logic lives in
  commands/delegate.md, which is injected by the slash command on
  demand. Update README.md and CLAUDE.md references accordingly.

Qodo #5 (team_mode_score fallback) and #6 (team_mode_active auto-create)
are intentional / false positives and will be replied to on the PR.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: barkain

CLAUDE.md

@@ -171,7 +171,7 @@ Two team workflow patterns:
 - Agents use conditional COMMUNICATION MODE (teammate messaging vs `DONE|{path}`)
 - State files: `.claude/state/team_mode_active`, `.claude/state/team_config.json`
 
-Agent selection uses keyword matching (≥2 matches threshold, highest count wins). Falls back to general-purpose if 0-1 matches. See `system-prompts/workflow_orchestrator.md` for keyword lists.
+Agent selection uses keyword matching (≥2 matches threshold, highest count wins). Falls back to general-purpose if 0-1 matches. See `commands/delegate.md` (Available Specialized Agents section) for keyword lists.
 
 Agent config format: YAML frontmatter (`name`, `description`, optional `tools`/`model`/`color`) + markdown system prompt body. All agents enforce `DONE|{output_file}` return format. Custom agent instructions include:
 - Return format must be exactly `DONE|{output_file_path}` (no summaries or explanations)
@@ -270,7 +270,7 @@ ls .claude/state/delegation_active 2>/dev/null  # delegation in progress?
 ```
 
 **Multi-step not detected:**
-Ensure SessionStart hooks are installed (inject_workflow_orchestrator.py) so that workflow_orchestrator.md is injected and native plan mode (EnterPlanMode/ExitPlanMode) is available. Use connectors in prompts: "and then", "with", "including".
+Ensure the SessionStart hook is installed (`hooks/SessionStart/inject_all.py`) so `orchestrator_stub.md` is injected and the main agent knows to route multi-step work through `/workflow-orchestrator:delegate`. Use connectors in prompts: "and then", "with", "including".
 
 **TeamCreate blocked:**
 ```bash

README.md

@@ -272,14 +272,14 @@ The `plugin-hooks.json` configures the delegation enforcement hooks using cross-
 The `inject_all.py` hook consolidates 3 SessionStart hooks into 1 Python script:
 
 **On startup/resume (all sessions):**
-- Injects orchestrator stub (`orchestrator_stub.md`, ~1.1KB): registers `/workflow-orchestrator:delegate` and `/workflow-orchestrator:bypass` commands. Minimal overhead (~200 tokens).
+- Injects orchestrator routing stub (`orchestrator_stub.md`, ~1.1KB): points the main agent at `/workflow-orchestrator:delegate` for multi-step work. Minimal overhead (~200 tokens).
 - Optionally injects token-efficient CLI guide (`token_efficient_cli.md`, ~1.9KB, gated by `CLAUDE_TOKEN_EFFICIENCY=1` env var). Teaches compact flags and command patterns.
 - Output style loaded natively from plugin.json `outputStyles` field (no injection required). Saves ~1.5K tokens.
 
 **On first delegation (lazy load):**
-- Full `workflow_orchestrator.md` (~7.5KB) loaded only when `/workflow-orchestrator:delegate` runs or multi-step detected.
+- Full orchestrator logic (planning instructions, agent catalog, wave scheduling, team-mode execution) lives inline in `commands/delegate.md` and is loaded only when `/workflow-orchestrator:delegate` runs.
 
-**Net savings:** ~6.6K tokens off session startup. Sessions pay only for orchestrator when delegation is used.
+**Net savings:** ~6.6K tokens off session startup. Sessions pay the orchestration tax only when delegation is actually used.
 
 **What this enables:**
 - Multi-step task detection via pattern matching
@@ -347,9 +347,9 @@ The `/workflow-orchestrator:delegate` command provides intelligent task delegati
 6. Creates task list via TaskCreate
 7. Exits plan mode (ExitPlanMode) and executes phases as directed by the plan
 
-### 4. Workflow Orchestration System Prompt (`system-prompts/workflow_orchestrator.md`)
+### 4. Orchestrator Command (`commands/delegate.md`)
 
-Enables multi-step workflow detection and preparation for complex tasks. Works in conjunction with native plan mode (EnterPlanMode/ExitPlanMode).
+The `/workflow-orchestrator:delegate` slash command loads the full orchestrator logic on demand: multi-step detection, plan-mode instructions, agent catalog, wave scheduling, and team-mode execution. Works in conjunction with native plan mode (EnterPlanMode/ExitPlanMode).
 
 **Activate via:**
 Simply start a Claude code session

hooks/PostToolUse/remind_skill_continuation.py

@@ -16,18 +16,19 @@
 import logging
 import os
 import sys
+import tempfile
 from pathlib import Path
 
 # Force UTF-8 output on Windows (fixes emoji encoding errors)
 if sys.platform == "win32":
     sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding="utf-8", errors="replace")
     sys.stderr = io.TextIOWrapper(sys.stderr.buffer, encoding="utf-8", errors="replace")
 
-# Setup debug logging
+# Setup debug logging (cross-platform temp path)
 DEBUG = os.environ.get("DEBUG_DELEGATION_HOOK", "0") == "1"
 if DEBUG:
     logging.basicConfig(
-        filename="/tmp/delegation_hook_debug.log",  # noqa: S108
+        filename=str(Path(tempfile.gettempdir()) / "delegation_hook_debug.log"),
         level=logging.DEBUG,
         format="%(asctime)s - remind_skill_continuation - %(message)s",
     )
@@ -65,7 +66,7 @@ def _create_continuation_state(reason: str) -> None:
             "additionalContext": CONTINUATION_CONTEXT,
         }
     }
-    print(json.dumps(output, ensure_ascii=False))  # noqa: T201 — hook stdout JSON
+    sys.stdout.write(json.dumps(output, ensure_ascii=False) + "\n")
 
 
 def _zero_violations_counter() -> None:

hooks/PostToolUse/validate_task_graph_depth.py

@@ -1,4 +1,7 @@
 #!/usr/bin/env python3
+# /// script
+# requires-python = ">=3.12"
+# ///
 """
 PostToolUse Hook: Task Graph Depth Hint (cross-platform)
 

hooks/PreToolUse/require_delegation.py

@@ -150,5 +150,8 @@ def main() -> int:
     try:
         sys.exit(main())
     except Exception as e:  # noqa: BLE001
+        # Surface internal errors per hook exit-code policy (1 = hook error).
+        # Exit 1 is a non-blocking error in Claude Code — the tool call still
+        # proceeds, but the failure is visible in debug output.
         logger.error("require_delegation hook error: %s", e)
-        sys.exit(0)  # Never block, even on internal errors
+        sys.exit(1)

hooks/PreToolUse/validate_task_graph_compliance.py

@@ -1,4 +1,7 @@
 #!/usr/bin/env python3
+# /// script
+# requires-python = ">=3.12"
+# ///
 """
 PreToolUse Hook: Task Graph Compliance Hint (cross-platform)