Context Portal MCP Server

Project memory bank providing structured SQLite storage and semantic search for AI agents.

Data last scanned today · Reviewed today

39 4 dangerous tools29 tools★ 764PythonApache-2.0

Overview

Context Portal (ConPort) functions as a dedicated memory bank for software projects. It uses a SQLite-based database per workspace to manage architectural decisions, system designs, and task progress, providing a structured alternative to standard file-based context management. The server leverages vector embeddings to enable semantic search, allowing AI assistants to perform Retrieval Augmented Generation (RAG). By maintaining explicit relationships between project entities, it creates a searchable knowledge graph that improves the accuracy of context-aware responses within IDEs.

Our verdict

This server carries a high-risk rating with a security score of 39/100. The score is primarily impacted by the fact that the project is community-maintained and lacks editorial review for critical components like authentication methods and read-only mode implementation. Due to the lack of oversight on security implementations, this tool should be used with caution in environments where access control and data sensitivity are prioritized.

Uses SQLite per workspace for structured local storage
Supports semantic search via vector data storage
Builds a project-specific knowledge graph for context
Zero direct dependencies reduce supply-chain surface area

High risk score due to lack of editorial security review
Last commit was 182 days ago
Authentication and read-only mechanisms are not yet reviewed

Tools

Tool	Description	Risk
get_workspace_detection_info	Exposes a diagnostic dictionary showing workspace detection information.	low
get_product_context	Retrieves the overall project goals, features, and architecture.	low
update_product_context	Updates the product context.	medium
get_active_context	Retrieves the current working focus, recent changes, and open issues.	low
update_active_context	Updates the active context.	medium
log_decision	Logs an architectural or implementation decision.	medium
get_decisions	Retrieves logged decisions.	low
search_decisions_fts	Full-text search across decision fields.	low
delete_decision_by_id	Deletes a decision by its ID.	high
log_progress	Logs a progress entry or task status.	medium

Compatibility

Client	Local	Docker	Remote	Read-only
ChatGPT
Claude Desktop
Cursor
VS Code
Windsurf

Frequently asked questions

›What software prerequisites are required to run the Context Portal MCP server?

You need Python version 3.8 or higher installed and added to your system's PATH. The use of uv is highly recommended as a package manager to simplify environment setup.

›How does this server handle project data persistence?

Context Portal uses a SQLite database backend to store project context, creating one database file per workspace to ensure structured and queryable memory.

›What is the recommended method for configuring the server in an IDE?

The recommended approach is to use uvx in your MCP client settings to execute the package directly from PyPI without managing manual virtual environments.

›How should I handle the workspace_id configuration?

You can either provide an absolute path for the workspace_id at launch or omit it to allow the server to initialize the database upon the first tool call.

›Does the server provide logging capabilities?

Yes, you can specify a file path using the --log-file argument and set the verbosity using --log-level, which defaults to INFO if not specified.