claude-flow-novice 2.15.3 → 2.15.4

package/claude-assets/skills/agent-template-generator/generate-agent.sh

@@ -0,0 +1,405 @@
+#!/usr/bin/env bash
+#
+# Agent Template Generator
+# Creates new agent profiles with consistent structure and validation patterns
+#
+# Usage:
+#   ./generate-agent.sh --name <agent-name> --type <agent-type> --description <description>
+#
+# Example:
+#   ./generate-agent.sh \
+#     --name "api-security-specialist" \
+#     --type "specialist" \
+#     --description "MUST BE USED for API security audits and penetration testing"
+#
+# Agent Types:
+#   - specialist: Domain-specific implementation agent (Loop 3)
+#   - validator: Code review and quality validation agent (Loop 2)
+#   - coordinator: Multi-agent orchestration and planning agent
+#   - utility: Supporting tools and helper agents
+
+set -euo pipefail
+
+# Default values
+AGENT_NAME=""
+AGENT_TYPE="specialist"
+AGENT_DESCRIPTION=""
+AGENT_TOOLS="[Read, Write, Edit, Bash, Grep, Glob, TodoWrite]"
+AGENT_MODEL="sonnet"
+AGENT_ACL_LEVEL="1"
+OUTPUT_DIR=""
+PROVIDER="zai"
+PROVIDER_MODEL="glm-4.6"
+
+# Parse command line arguments
+while [[ $# -gt 0 ]]; do
+    case $1 in
+        --name)
+            AGENT_NAME="$2"
+            shift 2
+            ;;
+        --type)
+            AGENT_TYPE="$2"
+            shift 2
+            ;;
+        --description)
+            AGENT_DESCRIPTION="$2"
+            shift 2
+            ;;
+        --tools)
+            AGENT_TOOLS="$2"
+            shift 2
+            ;;
+        --model)
+            AGENT_MODEL="$2"
+            shift 2
+            ;;
+        --acl-level)
+            AGENT_ACL_LEVEL="$2"
+            shift 2
+            ;;
+        --output-dir)
+            OUTPUT_DIR="$2"
+            shift 2
+            ;;
+        --provider)
+            PROVIDER="$2"
+            shift 2
+            ;;
+        --provider-model)
+            PROVIDER_MODEL="$2"
+            shift 2
+            ;;
+        --help)
+            echo "Usage: $0 --name <agent-name> --type <agent-type> --description <description>"
+            echo ""
+            echo "Required arguments:"
+            echo "  --name              Agent name (lowercase with hyphens)"
+            echo "  --description       Agent description (MUST BE USED pattern)"
+            echo ""
+            echo "Optional arguments:"
+            echo "  --type              Agent type (specialist|validator|coordinator|utility) [default: specialist]"
+            echo "  --tools             Tool list in JSON array format [default: [Read, Write, Edit, Bash, Grep, Glob, TodoWrite]]"
+            echo "  --model             Claude model (sonnet|opus|haiku) [default: sonnet]"
+            echo "  --acl-level         Access control level 1-3 [default: 1]"
+            echo "  --output-dir        Output directory [default: .claude/agents/cfn-dev-team/<type>s/]"
+            echo "  --provider          AI provider (zai|kimi|openrouter|anthropic) [default: zai]"
+            echo "  --provider-model    Provider-specific model [default: glm-4.6]"
+            exit 0
+            ;;
+        *)
+            echo "Unknown option: $1"
+            echo "Use --help for usage information"
+            exit 1
+            ;;
+    esac
+done
+
+# Validate required arguments
+if [[ -z "$AGENT_NAME" ]]; then
+    echo "Error: --name is required"
+    exit 1
+fi
+
+if [[ -z "$AGENT_DESCRIPTION" ]]; then
+    echo "Error: --description is required"
+    exit 1
+fi
+
+# Validate AGENT_MODEL (must be sonnet, opus, or haiku)
+if [[ ! "$AGENT_MODEL" =~ ^(sonnet|opus|haiku)$ ]]; then
+    echo "❌ ERROR: Invalid model specified: '$AGENT_MODEL'" >&2
+    echo "   Allowed values: sonnet, opus, haiku" >&2
+    exit 1
+fi
+
+# Validate AGENT_ACL_LEVEL (must be 1, 2, or 3)
+if [[ ! "$AGENT_ACL_LEVEL" =~ ^[1-3]$ ]]; then
+    echo "❌ ERROR: Invalid ACL level specified: '$AGENT_ACL_LEVEL'" >&2
+    echo "   Allowed values: 1, 2, 3" >&2
+    exit 1
+fi
+
+# Validate AGENT_TOOLS (must be valid JSON array)
+if command -v jq >/dev/null 2>&1; then
+    if ! echo "$AGENT_TOOLS" | jq -e 'if type == "array" then true else false end' >/dev/null 2>&1; then
+        echo "❌ ERROR: Invalid tools format: '$AGENT_TOOLS'" >&2
+        echo "   Expected: JSON array of tool names" >&2
+        echo "   Example: '[Read, Write, Edit, Bash]'" >&2
+        exit 1
+    fi
+else
+    # Fallback: basic bracket check if jq not available
+    if [[ ! "$AGENT_TOOLS" =~ ^\[.*\]$ ]]; then
+        echo "❌ ERROR: Invalid tools format: '$AGENT_TOOLS'" >&2
+        echo "   Expected: JSON array of tool names" >&2
+        echo "   Example: '[Read, Write, Edit, Bash]'" >&2
+        echo "   Note: jq not available for full JSON validation" >&2
+        exit 1
+    fi
+fi
+
+# Determine output directory based on agent type
+if [[ -z "$OUTPUT_DIR" ]]; then
+    case "$AGENT_TYPE" in
+        specialist)
+            OUTPUT_DIR=".claude/agents/cfn-dev-team/developers"
+            ;;
+        validator)
+            OUTPUT_DIR=".claude/agents/cfn-dev-team/reviewers/quality"
+            ;;
+        coordinator)
+            OUTPUT_DIR=".claude/agents/cfn-dev-team/coordinators"
+            ;;
+        utility)
+            OUTPUT_DIR=".claude/agents/cfn-dev-team/utility"
+            ;;
+        *)
+            echo "Error: Invalid agent type: $AGENT_TYPE"
+            echo "Valid types: specialist, validator, coordinator, utility"
+            exit 1
+            ;;
+    esac
+fi
+
+# Create output directory if it doesn't exist
+mkdir -p "$OUTPUT_DIR"
+
+# Output file path
+OUTPUT_FILE="$OUTPUT_DIR/${AGENT_NAME}.md"
+
+# Check if file already exists
+if [[ -f "$OUTPUT_FILE" ]]; then
+    echo "Error: Agent file already exists: $OUTPUT_FILE"
+    echo "Remove existing file or choose a different name"
+    exit 1
+fi
+
+# Generate agent template
+cat > "$OUTPUT_FILE" << EOF
+---
+name: $AGENT_NAME
+description: $AGENT_DESCRIPTION
+tools: $AGENT_TOOLS
+model: $AGENT_MODEL
+type: $AGENT_TYPE
+acl_level: $AGENT_ACL_LEVEL
+validation_hooks:
+  - agent-template-validator
+  - test-coverage-validator
+---
+
+<!-- PROVIDER_PARAMETERS
+provider: $PROVIDER
+model: $PROVIDER_MODEL
+-->
+
+# ${AGENT_NAME^} Agent
+
+You are a ${AGENT_TYPE} agent specialized in [DOMAIN EXPERTISE - TO BE CUSTOMIZED].
+
+## Core Responsibilities
+
+[CUSTOMIZE: List 3-5 primary responsibilities]
+
+1. [Responsibility 1]
+2. [Responsibility 2]
+3. [Responsibility 3]
+
+---
+
+## Success Criteria Integration (REQUIRED)
+
+### Validation Pattern
+
+All agents MUST use the centralized JSON validation skill:
+
+\`\`\`bash
+# Source centralized validation skill
+source .claude/skills/json-validation/validate-success-criteria.sh
+
+# Validate on startup (exits on invalid JSON)
+validate_success_criteria || exit 1
+
+# Access parsed criteria
+if [[ -n "\${CRITERIA:-}" ]]; then
+    # List all test suites
+    list_test_suites
+
+    # Get specific test command
+    TEST_CMD=\$(get_test_command "unit-tests")
+
+    # Get pass threshold
+    THRESHOLD=\$(get_pass_threshold "unit-tests")
+fi
+\`\`\`
+
+### TDD Protocol (Test-Driven Development)
+
+**Write Tests First:**
+- Extract test requirements from success criteria
+- Write failing tests for each requirement
+- Ensure test coverage ≥80%
+- *Guidance: Typically ~15-20 min, adjust based on complexity*
+
+**Implement:**
+- Write minimum code to pass tests
+- Run tests continuously (\`npm test --watch\` or framework equivalent)
+- Refactor for quality
+- *Guidance: Typically ~30-40 min, adjust based on complexity*
+
+**Validate:**
+- Run full test suite from success criteria
+- Verify pass rate meets threshold (Standard: ≥95%)
+- Check coverage metrics
+- *Guidance: Typically ~5 min for validation*
+
+---
+
+## Test-Driven Validation (NOT Confidence Scores)
+
+### Execute Tests
+
+\`\`\`bash
+# Get test command from success criteria
+TEST_CMD=\$(get_test_command "primary-test-suite")
+
+if [[ -n "\$TEST_CMD" ]]; then
+    # Execute tests and capture output
+    TEST_OUTPUT=\$(eval "\$TEST_CMD" 2>&1)
+
+    # Parse test results using CFN helper
+    RESULTS=\$(./.claude/skills/cfn-loop-orchestration/helpers/parse-test-results.sh \\
+        "jest" "\$TEST_OUTPUT")
+
+    # Report test results (NOT confidence scores)
+    ./.claude/skills/cfn-redis-coordination/report-completion.sh \\
+        --task-id "\$TASK_ID" \\
+        --agent-id "\$AGENT_ID" \\
+        --test-results "\$RESULTS"
+else
+    echo "⚠️  No test command found in success criteria"
+    echo "   Proceeding without test validation"
+fi
+\`\`\`
+
+### Completion Protocol
+
+**DO NOT** report subjective confidence scores (0.0-1.0).
+
+**DO** report objective test metrics:
+
+\`\`\`bash
+echo "Test Results:"
+echo "  Test Suite: [suite-name]"
+echo "  Total Tests: [count]"
+echo "  Passed: [count]"
+echo "  Failed: [count]"
+echo "  Pass Rate: [percentage]%"
+echo "  Threshold: [required-rate]%"
+echo "  Status: [PASS|FAIL]"
+\`\`\`
+
+---
+
+## Domain-Specific Guidelines
+
+[CUSTOMIZE: Add domain-specific best practices, patterns, anti-patterns]
+
+### Best Practices
+
+1. [Practice 1]
+2. [Practice 2]
+3. [Practice 3]
+
+### Common Pitfalls
+
+1. [Pitfall 1]
+2. [Pitfall 2]
+3. [Pitfall 3]
+
+### Recommended Tools
+
+- [Tool 1]: [Use case]
+- [Tool 2]: [Use case]
+- [Tool 3]: [Use case]
+
+---
+
+## Success Metrics
+
+[CUSTOMIZE: Define objective success metrics]
+
+- [Metric 1]: [Target]
+- [Metric 2]: [Target]
+- [Metric 3]: [Target]
+- Overall test pass rate: ≥ 0.95 (Standard mode)
+
+---
+
+## Example Workflows
+
+### Workflow 1: [Scenario Name]
+
+\`\`\`bash
+# Step 1: [Description]
+[command or action]
+
+# Step 2: [Description]
+[command or action]
+
+# Step 3: [Description]
+[command or action]
+\`\`\`
+
+### Workflow 2: [Scenario Name]
+
+\`\`\`bash
+# Step 1: [Description]
+[command or action]
+
+# Step 2: [Description]
+[command or action]
+\`\`\`
+
+---
+
+## Integration Points
+
+[CUSTOMIZE: Document how this agent integrates with other agents]
+
+- **Loop 3 (Implementation)**: [How this agent participates]
+- **Loop 2 (Validation)**: [How this agent validates]
+- **Dependencies**: [Other agents or services required]
+
+---
+
+## References
+
+[CUSTOMIZE: Add relevant documentation links]
+
+- Internal Docs: [Link]
+- External Docs: [Link]
+- Related Agents: [List]
+
+---
+
+**Status:** Generated template - requires customization
+**Version:** 1.0.0
+**Created:** $(date +%Y-%m-%d)
+EOF
+
+echo "✅ Agent template created: $OUTPUT_FILE"
+echo ""
+echo "Next steps:"
+echo "1. Customize the [DOMAIN EXPERTISE] placeholder"
+echo "2. Fill in Core Responsibilities"
+echo "3. Customize Domain-Specific Guidelines"
+echo "4. Define Success Metrics"
+echo "5. Add Example Workflows"
+echo "6. Document Integration Points"
+echo "7. Add relevant references"
+echo ""
+echo "Validation Pattern: ✅ Automatically included"
+echo "Test-Driven Protocol: ✅ Automatically included"
+echo "Provider Configuration: ✅ Set to $PROVIDER ($PROVIDER_MODEL)"