Fix 4 codex-identified bugs

.github/workflows/create-release-tags.yml

@@ -38,11 +38,6 @@ jobs:
           git config user.name "github-actions[bot]"
           git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
 
-      - name: Pull latest changes
-        run: |
-          # Ensure we have the version bump commit
-          git pull origin main
-
       - name: Create tags
         id: tags
         run: |
@@ -66,11 +61,17 @@ jobs:
               continue
             fi
 
-            # Create annotated tag
-            git tag -a "$TAG" -m "Release ${plugin} v${VERSION}
+            # Create annotated tag on the version bump commit (not HEAD which may have moved)
+            COMMIT_SHA="${{ github.event.client_payload.commit_sha }}"
+            if [ -z "$COMMIT_SHA" ]; then
+              echo "::error::commit_sha not provided in payload"
+              exit 1
+            fi
+
+            git tag -a "$TAG" "$COMMIT_SHA" -m "Release ${plugin} v${VERSION}
 
 Auto-generated by GitHub Actions workflow.
-Trigger commit: ${{ github.event.client_payload.trigger_sha }}"
+Version bump commit: $COMMIT_SHA"
 
             echo "Created tag: $TAG"
             CREATED_TAGS+=("$TAG")
@@ -115,4 +116,5 @@ Trigger commit: ${{ github.event.client_payload.trigger_sha }}"
 
           echo "" >> $GITHUB_STEP_SUMMARY
           echo "---" >> $GITHUB_STEP_SUMMARY
-          echo "**Trigger commit:** \`${{ github.event.client_payload.trigger_sha }}\`" >> $GITHUB_STEP_SUMMARY
+          echo "**Version bump commit (tagged):** \`${{ github.event.client_payload.commit_sha }}\`" >> $GITHUB_STEP_SUMMARY
+          echo "**Original trigger commit:** \`${{ github.event.client_payload.trigger_sha }}\`" >> $GITHUB_STEP_SUMMARY

.github/workflows/detect-plugin-changes.yml

@@ -53,7 +53,7 @@ jobs:
       - name: Checkout repository
         uses: actions/checkout@v4
         with:
-          fetch-depth: 2  # Need HEAD~1 for diff comparison
+          fetch-depth: 0  # Full history for multi-commit push detection
 
       - name: Detect changed plugins
         id: changes
@@ -63,8 +63,13 @@ jobs:
           # Local plugins to check (excludes external golang-workflow)
           LOCAL_PLUGINS=("security-hooks" "todo-log" "version-control" "bash-workflow" "claude-code-guide" "slash-command-guide")
 
-          # Get changed files between HEAD and HEAD~1
-          CHANGED_FILES=$(git diff --name-only HEAD~1 HEAD 2>/dev/null || git diff --name-only HEAD)
+          # Get changed files since push started (handles multi-commit pushes)
+          if [ "${{ github.event.before }}" != "0000000000000000000000000000000000000000" ]; then
+            CHANGED_FILES=$(git diff --name-only "${{ github.event.before }}" HEAD)
+          else
+            # Initial push or force push - compare with HEAD~1 as fallback
+            CHANGED_FILES=$(git diff --name-only HEAD~1 HEAD 2>/dev/null || git diff --name-only HEAD)
+          fi
           echo "Changed files:"
           echo "$CHANGED_FILES"
 

CLAUDE.md

@@ -33,6 +33,9 @@ Claude Code plugin marketplace with seven plugins:
 /plugin list
 /hooks
 claude --debug
+
+# Set GitHub secret from current auth token (no intermediate files)
+gh auth token | gh secret set SECRET_NAME
 ```
 
 ### Development Workflow

bash-workflow/agents/plan-devops.md

@@ -0,0 +1,21 @@
+---
+name: plan-devops
+description: "Analyze operational considerations for bash tasks. Focuses on portability, CI/CD, and idempotency."
+model: haiku
+color: purple
+---
+
+Analyze operational requirements for the given bash task.
+
+## Focus Areas
+
+1. **Portability** - macOS vs Linux, BSD vs GNU, path differences
+2. **CI/CD** - Exit codes, output parsing, env expectations
+3. **Deployment** - Installation, dependencies, upgrades
+4. **Idempotency** - Safe reruns, state management, cleanup
+
+## Output
+
+Write to `~/.claude/bash-workflow/plan-devops.md`:
+- Platform compatibility notes
+- CI/CD and idempotency strategy

bash-workflow/agents/plan-implementation.md

@@ -0,0 +1,22 @@
+---
+name: plan-implementation
+description: "Design implementation approach for bash tasks. Focuses on script structure, function decomposition, and dependencies."
+model: haiku
+color: blue
+---
+
+Design the implementation approach for the given bash task.
+
+## Focus Areas
+
+1. **Script Structure** - Entry points, module separation, config handling
+2. **Function Decomposition** - Single responsibility, reusable utilities
+3. **Dependencies** - Implementation order, shared resources
+4. **Codebase Patterns** - Existing conventions, error/logging patterns
+
+## Output
+
+Write to `~/.claude/bash-workflow/plan-implementation.md`:
+- Proposed file structure
+- Key functions and responsibilities
+- Implementation order

bash-workflow/agents/plan-security.md

@@ -0,0 +1,21 @@
+---
+name: plan-security
+description: "Design security considerations for bash tasks. Focuses on input validation, permissions, and secrets."
+model: haiku
+color: orange
+---
+
+Identify security considerations for the given bash task.
+
+## Focus Areas
+
+1. **Input Validation** - Sanitization, path traversal, injection prevention
+2. **Permissions** - File modes, sudo usage, umask
+3. **Secrets** - No hardcoding, env vars, secure storage
+4. **Unsafe Patterns** - eval, unquoted vars, glob risks, temp files
+
+## Output
+
+Write to `~/.claude/bash-workflow/plan-security.md`:
+- Input validation requirements
+- Permission model and secret handling

bash-workflow/agents/plan-testing.md

@@ -0,0 +1,21 @@
+---
+name: plan-testing
+description: "Design test strategy for bash tasks. Focuses on bats tests, edge cases, and failure modes."
+model: haiku
+color: green
+---
+
+Design comprehensive test coverage for the given bash task.
+
+## Focus Areas
+
+1. **Behaviors** - Happy path, error handling, edge cases
+2. **Edge Cases** - Empty input, missing files, bad permissions, signals
+3. **Failure Modes** - Network, disk full, permission denied
+4. **Platforms** - macOS vs Linux, BSD vs GNU tools
+
+## Output
+
+Write to `~/.claude/bash-workflow/plan-testing.md`:
+- Test scenarios by feature
+- Edge cases and failure modes

security-hooks/scripts/check_secrets.py

@@ -413,7 +413,7 @@ def _run_secret_check(project_root: Path) -> None:
             }
         }
         print(json.dumps(output))
-        sys.exit(ExitCode.SUCCESS)  # Use exit 0 with JSON output
+        sys.exit(ExitCode.BLOCKED)  # Exit 2 ensures Claude sees the deny decision
 
     sys.exit(ExitCode.SUCCESS)
 

security-hooks/scripts/test_check_secrets.py

@@ -1135,8 +1135,8 @@ def mock_subprocess_run(
                     with pytest.raises(SystemExit) as exc_info:
                         main()
 
-        # With JSON output, exit code is 0 (success) but with deny decision
-        assert exc_info.value.code == ExitCode.SUCCESS
+        # Exit code 2 (BLOCKED) ensures Claude respects the deny decision
+        assert exc_info.value.code == ExitCode.BLOCKED
 
         # Verify JSON output structure
         captured = capsys.readouterr()