Skip to main content
Glama
SECURITY.md4.55 kB
# Security Policy ## Supported Versions | Version | Supported | | ------- | ------------------ | | 1.0.x | :white_check_mark: | | < 1.0 | :x: | ## Reporting a Vulnerability We take security seriously. If you discover a security vulnerability in AIStack-MCP, please report it responsibly. ### How to Report 1. **DO NOT** open a public GitHub issue for security vulnerabilities 2. Email security concerns to the maintainers directly 3. Include: - Description of the vulnerability - Steps to reproduce - Potential impact - Suggested fix (if any) ### Response Timeline - **Initial Response**: Within 48 hours - **Status Update**: Within 7 days - **Resolution Target**: Within 30 days (severity-dependent) ### What to Expect 1. Acknowledgment of your report 2. Assessment of the vulnerability 3. Development of a fix 4. Coordinated disclosure (if applicable) 5. Credit in release notes (unless you prefer anonymity) --- ## Security Best Practices ### MCP Server Security #### Single-Repo Mode (Maximum Security) ```powershell # Ensures complete isolation .\scripts\switch_to_single_repo.ps1 ``` - ✅ Each workspace has isolated MCP servers - ✅ Explicit permissions per repository - ✅ No cross-repository access - ✅ Portable `${workspaceFolder}` configuration #### Multi-Repo Mode (Controlled Access) ```powershell # Only link trusted repositories .\scripts\switch_to_multi_repo.ps1 ``` - ⚠️ CORE workspace has access to ALL linked repositories - ⚠️ Only link repositories with compatible security levels - ✅ Explicit linking required (no automatic discovery) - ✅ Mode indicator tracks configuration state ### Data Privacy #### Local Processing AIStack-MCP is designed with local-first principles: | Operation | Where Processed | Data Leaves Machine? | |-----------|-----------------|---------------------| | Semantic search | Qdrant (local) | ❌ No | | Pattern analysis | Ollama (local) | ❌ No | | Context compression | Ollama (local) | ❌ No | | Code generation | Claude (cloud) | ✅ Compressed context only | #### Sensitive Data Guidelines ``` ❌ DON'T: - Link production credential repositories in multi-repo mode - Store API keys in indexed files - Index .env or secrets files ✅ DO: - Use single-repo mode for sensitive repositories - Keep secrets in .gitignore - Review linked repos before enabling multi-repo mode ``` ### Configuration Security #### Validate Configurations ```powershell # Always validate after changes python scripts\validate_mcp_config.py --strict ``` #### Audit Mode Changes Check `.cursor/ACTIVE_MODE.txt` to verify: - Current mode (single-repo vs multi-repo) - Linked repositories (in multi-repo mode) - Last update timestamp ### Network Security #### Service Ports | Service | Default Port | Recommendation | |---------|--------------|----------------| | Ollama | 11434 | Bind to localhost only | | Qdrant | 6333 | Bind to localhost only | #### Firewall Configuration ```powershell # Ollama and Qdrant should only be accessible locally # Default configurations already bind to localhost # Do NOT expose these ports to external networks ``` ### CI/CD Security #### Validation in Pipelines ```yaml # Prevent insecure configurations from being committed - name: Security Validation run: python scripts/validate_mcp_config.py --strict ``` #### Secrets Management - Never commit `.cursor/mcp.json` with hardcoded tokens - Use environment variables for sensitive values - GitHub tokens should use minimal required permissions --- ## Known Security Considerations ### Windows Symlinks Creating symlinks on Windows requires Administrator privileges. This is a Windows security feature, not a vulnerability. **Alternatives:** - Use `-Clone` flag to clone instead of symlink - Run PowerShell as Administrator when needed ### MCP Server Trust MCP servers have access to files within their configured directories. This is by design. **Mitigations:** - Single-repo mode limits access to one repository - Multi-repo mode requires explicit linking - Never link untrusted repositories --- ## Security Changelog ### v1.0.0 - Implemented workspace isolation pattern - Added mode indicator for audit trail - Added configuration validation scripts - Documented security best practices --- ## Contact For security-related inquiries, please reach out through: - GitHub Security Advisory (for vulnerabilities) - GitHub Issues (for security questions, non-sensitive) Thank you for helping keep AIStack-MCP secure!

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/mjdevaccount/AIStack-MCP'

If you have feedback or need assistance with the MCP directory API, please join our Discord server