security: Restrict Claude and improve Dependabot handling

[niksacdev]

Summary

• Restrict Claude AI assistant to repository owner only
• Fix Dependabot PR handling in workflows
• Create GitHub issues for future enhancements

Changes

Security Improvements

• Modified to only allow @niksacdev to trigger Claude
• Added to notify users about restrictions
• Updated with AI assistant usage policy (v1.1)

Workflow Fixes

• Skip Claude Code Review for Dependabot PRs (they don't have access to secrets)
• Added dedicated workflow
• Auto-approve and merge Dependabot version updates

Documentation

• Created 10 GitHub issues (feat: Add Document OCR MCP Server for Intake Agent #19-feat: Implement Appeals Agent for Decision Review #28) for future MCP servers and agent enhancements
• Removed file in favor of actual GitHub issues

Test Plan

• Claude restrictions tested locally
• Workflow changes validated
• Dependabot auto-merge to be tested when PRs are recreated

🤖 Generated with Claude Code

[github-actions]

🔬 Expert Engineer Review Analysis

Target Branch: main
Changes Analysis:

📁 File Changes

• 📝 Modified: .github/workflows/claude-code-review.yml
• ➕ Added: .github/workflows/claude-restricted-message.yml
• 📝 Modified: .github/workflows/claude.yml
• ➕ Added: .github/workflows/dependabot-auto-merge.yml
• 📝 Modified: CLAUDE.md
• 📝 Modified: SECURITY.md
• ❌ Deleted: github_issues.md

🧪 Test Coverage Analysis

❌ Test failures detected - Review required

ImportError while importing test module '/home/runner/work/multi-agent-system/multi-agent-system/tests/test_agent_registry.py'.
Hint: make sure your test modules/packages have valid Python names.
Traceback:
/opt/hostedtoolcache/Python/3.10.18/x64/lib/python3.10/importlib/__init__.py:126: in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
tests/test_agent_registry.py:10: in <module>
    from agents import Agent
E   ImportError: cannot import name 'Agent' from 'agents' (/home/runner/work/multi-agent-system/multi-agent-system/loan_processing/agents/__init__.py)
------------------------------- Captured stdout --------------------------------
📝 Using console logging (set AZURE_MONITOR_CONNECTION_STRING for Azure integration)
=============================== warnings summary ===============================
.venv/lib/python3.10/site-packages/pydantic/_internal/_generate_schema.py:298: 10 warnings
  /home/runner/work/multi-agent-system/multi-agent-system/.venv/lib/python3.10/site-packages/pydantic/_internal/_generate_schema.py:298: PydanticDeprecatedSince20: `json_encoders` is deprecated. See https://docs.pydantic.dev/2.11/concepts/serialization/#custom-serializers for alternatives. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.11/migration/
    warnings.warn(

-- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html
=========================== short test summary info ============================
ERROR tests/test_agent_registry.py
!!!!!!!!!!!!!!!!!!!! Interrupted: 1 error during collection !!!!!!!!!!!!!!!!!!!!
======================== 10 warnings, 1 error in 0.31s =========================

🏗️ Architecture Impact Analysis

✅ Low Impact: No core architecture files modified

🔒 Security Analysis

⚠️ Potential hardcoded secrets detected

⚠️ Manual security review recommended

📊 Code Quality Metrics

⚠️ Linting: 4 issues found

View linting issues

error: invalid value 'text' for '--output-format <OUTPUT_FORMAT>'
  [possible values: concise, full, json, json-lines, junit, grouped, github, gitlab, pylint, rdjson, azure, sarif]

For more information, try '--help'.

✅ **Formatting:** Code properly formatted

🎯 Review Recommendations

3. 📋 General Checklist
   • All tests pass (✅ automated check)
   • Coverage ≥90% on core components (✅ automated check)
   • Code follows established patterns
   • Documentation updated if needed
   • Breaking changes documented

---

🤖 This review was automatically generated. Human expert review may still be required for complex changes.