Context Portal MCP Release Notes

v0.3.8 (2025-10-30)

Dependency Management

• Updated Dependencies: Resolved dependency conflicts by updating package version constraints
• FastMCP Security: Ensured compatibility with secure FastMCP versions (>=2.13.0)
• HTTPX Compatibility: Updated to require httpx>=0.28.1 to match FastMCP requirements

Installation

uvx --from git+https://github.com/GreatScottyMac/context-portal.git@v0.3.8 conport-mcp --mode stdio

v0.3.7 (2025-10-30)

Critical Fix

• Resolved FastAPI/Starlette Dependency Conflict: Fixed a dependency conflict that was preventing uvx installation. The issue occurred because fastapi==0.116.2 required starlette<0.49.0, while we needed starlette>=0.49.1 for the CVE-2025-62727 security fix.

Changes

• Updated FastAPI: Upgraded from 0.116.2 to >=0.119.1, which natively supports Starlette 0.49.1+
• Removed Explicit Starlette Dependency: No longer needed as FastAPI 0.119.1+ automatically includes the secure version of Starlette
• Maintained Security Posture: The update preserves all security fixes including CVE-2025-62727 (Starlette), CVE-2025-50181, and CVE-2025-50182 (urllib3)

Installation

uvx --from git+https://github.com/GreatScottyMac/context-portal.git conport-mcp --mode stdio

Or via pip:

pip install context-portal-mcp

v0.3.6 (2025-10-28)

Security

• Updated starlette to >=0.49.1 to remediate CVE-2025-62727 (High severity - O(n^2) DoS vulnerability in Range header merging in starlette.responses.FileResponse).
• Updated urllib3 to >=2.5.0 to remediate CVE-2025-50181 and CVE-2025-50182 (Moderate severity).

Packaging

• Updated project version to 0.3.6 in pyproject.toml.
• Added explicit urllib3>=2.5.0 dependency to both requirements.txt and pyproject.toml.

v0.3.5 (2025-10-22)

Security

• Bumped Authlib to ~=1.6.5 to remediate CVE-2025-61920 (High) and GHSA-g7f3-828f-7h7m (Moderate).
• Regenerated uv.lock to pin authlib==1.6.5 and align with current dependencies.
• Verified via full test run: 15 passed, 0 failed.

Packaging

• Updated project version to 0.3.5 in pyproject.toml.
• Ensured authlib is declared in pyproject.toml dependencies to keep locks and installs consistent across environments.

v0.3.4 (2025-09-18)

Critical Bug Fix

• String-to-Integer Coercion: Fixed a validation timing issue where field-level ge/le constraints in FastMCP tool definitions were preventing string-to-integer coercion from working properly. String parameters like "5" for limit were being rejected before the IntCoercionMixin could convert them to integers. The fix removes field-level constraints from 13 affected tools and replaces them with @model_validator(mode='after') methods in Pydantic models, ensuring coercion happens before validation.

Technical Details

Security

• Dependency hardening: Pin Authlib to ~=1.6.5 to address CVE-2025-61920 (High) and GHSA-g7f3-828f-7h7m (Moderate). Regenerated uv.lock to ensure 1.6.5 is locked. No runtime regressions observed (15/15 tests passing).
• Affected Tools: get_decisions, get_progress, get_system_patterns, get_custom_data, search_decisions_fts, search_custom_data_value_fts, search_project_glossary_fts, get_recent_activity_summary, semantic_search_conport, get_item_history, batch_log_items, delete_decision_by_id, delete_system_pattern_by_id
• Root Cause: FastMCP field-level ge=1 and le=25 constraints were applied before Pydantic model validation, preventing the custom IntCoercionMixin from converting string inputs to integers
• Solution: Moved all integer validation logic to @model_validator(mode='after') methods that run after field coercion

v0.3.3 (2025-09-18)

Fixes & Improvements

• Pydantic Validation Fix: Corrected an issue where Pydantic's ge=1 constraint was being applied before the IntCoercionMixin could convert string-based integers, causing validation errors. The fix ensures that string-to-integer coercion happens before validation, allowing for more flexible input.
• Timezone-aware datetimes: replaced naive UTC usage with aware UTC across models and DB code, and registered SQLite adapters/converters for reliable UTC round-tripping. Files: src/context_portal_mcp/db/models.py, src/context_portal_mcp/db/database.py.
• Integer-like string inputs: added lenient parsing that coerces digit-only strings to integers before validation in relevant argument models. Credit: @cipradu.
• Dependency security: addressed Starlette advisory GHSA-2c2j-9gv5-cj73 by upgrading FastAPI to 0.116.2 and constraining Starlette to >=0.47.2,<0.49.0; verified via pip-audit: "No known vulnerabilities found."

v0.3.0 (2025-09-08)

Features

• Universal Workspace Auto-Detection: Integrated multi-strategy workspace discovery (strong indicators, multiple indicators, existing context_portal/, environment variables, fallback). Eliminates need to hardcode --workspace_id in most MCP client configs. Includes new CLI flags: --auto-detect-workspace (default enabled), --no-auto-detect, and --workspace-search-start <path>.
• Diagnostic Tool: Added get_workspace_detection_info MCP tool to expose detection details for debugging ambiguous setups.
• Graceful ${workspaceFolder} Handling: If an IDE passes the literal ${workspaceFolder}, the server now warns and safely auto-detects instead of initializing incorrectly.
• Documentation: Added UNIVERSAL_WORKSPACE_DETECTION.md plus README section “Automatic Workspace Detection” with usage guidance and examples.

Notes

This release supersedes the stalled external contribution (original PR #60). Attribution preserved in commit metadata. Users are encouraged to remove hardcoded absolute paths where safe.

Upgrade Guidance

No migration steps required. Existing workflows with explicit --workspace_id continue to function. To leverage auto-detection, you may remove the flag (or allow per-call workspace_id injection).

v0.2.23 (2025-08-30)

Features

• Compact ConPort Strategy for Windsurf: Added a compact ConPort memory strategy file under 12k characters for Windsurf IDE compatibility, preserving core functionality while reducing size. (Credit: @kundeng, PR #55)
• Mem4Sprint Strategy and FTS5 Updates: Introduced mem4sprint strategy with flat categories, FTS5-safe examples, handler-only query normalization, and updated README for better IDE configuration. (Credit: @kundeng, PR #56)
• CLI Base Path Argument: Added --base-path CLI argument for custom storage locations, enhancing user flexibility. (Credit: @kundeng, PR #59)

Bug Fixes

• FTS5 Migration Guard: Wrapped FTS5 virtual table and trigger creation in try/except to gracefully degrade when unavailable, ensuring portable migrations. (Credit: @kundeng, PR #53)
• Deferred Workspace Resolution: Deferred workspace resolution when ${workspaceFolder} is literal to prevent mis-initialization for various MCP clients. (Credit: @kundeng, PR #54)
• Workspace ID Resolution Bugfix: Resolved workspace ID issue for better stability. (Credit: @yy1588133, PR #49)

v0.2.22 (2025-07-27)

Bug Fixes

• Startup Timeout: Resolved a client-side timeout issue by pre-warming the database connection on server startup. This ensures the server is fully initialized before accepting connections, preventing timeouts on the first tool call.

v0.2.21 (2025-07-24)

Features & Fixes

• Deferred Workspace Initialization: The server now initializes the workspace and database on the first tool call rather than on startup. This resolves a critical issue where an invalid or un-expanded ${workspaceFolder} variable from the client would prevent the server from starting. (Credit: @yy1588133)
• Full-Text Search (FTS): Introduced FTS5 virtual tables for decisions and custom_data. This significantly enhances search capabilities within ConPort. (Credit: @yy1588133)
• Comprehensive Test Suite: Added a new test script (test_conport_tools.py) to validate all ConPort MCP tools, improving stability and reliability. (Credit: @yy1588133)

v0.2.20 (2025-07-23)

Bug Fixes

Applied a fix to resolve the intermittent connection issue with the ConPort MCP server. A 500ms delay has been added to the startup sequence in src/context_portal_mcp/main.py to prevent a race condition.

v0.2.19 (2025-07-09)

Bug Fixes

• Pydantic Validation: Corrected a type coercion issue where Pydantic's gt=0 constraint for integer fields would fail when receiving numbers from the JSON-based MCP protocol. The validation has been updated to the more robust ge=1 (greater than or equal to 1) to ensure compatibility. (Credit: @akeech-chatham)

Version 0.2.18 - Hotfix for Log File Path Resolution

This release provides a hotfix to ensure the log file path is correctly resolved relative to the context_portal directory within the workspace, even when overridden by client-provided relative paths.

Key Fix:

• Log Path Resolution: The logging setup logic in main.py has been improved to robustly join the workspace_id, the context_portal directory, and any relative log file path. This corrects an issue where client-provided relative paths (e.g., ./logs/conport.log) were incorrectly resolved from the workspace root instead of from within the context_portal directory.

Impact:

This fix ensures that logs are always created in the intended context_portal/logs/ directory, improving consistency and organization.

Version 0.2.17 - Log File Path Standardization

This release changes the default location for the server's log file to a more standardized and predictable path within the workspace.

Key Change:

• Default Log Path: The --log-file argument now defaults to context_portal/logs/conport.log. The server will automatically create this directory structure within the active workspace if it doesn't exist. This keeps project-related files, including logs, neatly organized inside the context_portal directory.

Impact:

This change improves project organization by preventing log files from cluttering the root of the workspace.

Version 0.2.16 - IDE Compatibility Fix

This hotfix release addresses a critical startup failure when using the ConPort server with the latest versions of client IDEs (e.g., Roo Code v3.20.0 and later).

Key Fix:

• Logging System Refactor: The server's logging system has been refactored to prevent premature interaction with stdio streams. All logging is now configured after command-line arguments are parsed, resolving the OSError: [Errno 22] Invalid argument and ValueError: I/O operation on closed file errors that occurred during startup in stdio mode.

Impact:

This is a critical update for all users to ensure compatibility with the latest development tools. The server should now start and run reliably when launched from any IDE.

Version 0.2.14 - Database Stability and Bug Fixes

This release addresses a series of critical bugs related to database initialization and Alembic migrations, significantly improving the stability and reliability of the ConPort server.

Key Fixes:

• Database Initialization: Resolved a critical failure where new workspaces would fail to initialize the database due to a no such column: timestamp error. The root cause was a complex series of issues with Alembic migrations, which have now been fully repaired.
• Alembic Migration History: Repaired a corrupted Alembic migration history that was causing "Multiple head revisions are present" errors and preventing all database operations. The migration history is now clean and linear.
• get_system_patterns Tool: Fixed a bug where using the limit parameter would cause a Pydantic validation error. The tool now correctly handles the limit parameter.

Impact:

This is a critical stability release. All users should upgrade to ensure reliable database operation and to benefit from the bug fixes.

Version 0.2.11 - Cursor Initialization Hotfix

This release addresses a critical bug that caused NameError: name 'cursor' is not defined during database operations.

Key Fix:

• Cursor Initialization Fix: Corrected src/context_portal_mcp/db/database.py to ensure the cursor variable is always initialized before use within database functions. This resolves the NameError and ensures robust database interactions.

Impact:

This is a critical fix for server stability, ensuring all database operations function correctly.

Version 0.2.10 - Pydantic Model Hotfix

This release addresses a critical bug that caused MCP connection failures on Windows due to improper path escaping in the generated tool schema.

Key Fix:

• Pydantic Model Default Fix: Modified src/context_portal_mcp/db/models.py to add default=None to optional path arguments in the ExportConportToMarkdownArgs and ImportMarkdownToConportArgs models. This prevents Pydantic from generating default path values that contain unescaped backslashes on Windows, resolving the SyntaxError: Bad escaped character in JSON error during server startup.

Impact:

This is a critical fix for Windows users, ensuring the ConPort MCP server can start and run reliably.

Version 0.2.9 - Path Escaping Hotfix

This release addresses a critical bug that caused MCP connection failures on Windows due to improper path escaping.

Key Fix:

• JSON Escape Character Fix: Modified src/context_portal_mcp/db/database.py to use .as_posix() when constructing file paths for Alembic. This ensures that all paths use forward slashes, preventing SyntaxError: Bad escaped character in JSON errors when the server communicates with the client on Windows.

Impact:

This is a critical fix for Windows users, ensuring the ConPort MCP server can start and run reliably.

Version 0.2.8 - Alembic, Encoding, and Usability Enhancements

This release introduces several key improvements, including a fix for Alembic migrations, enhanced UTF-8 encoding for file operations, and a streamlined installation process.

Key Fixes & Enhancements:

• Alembic Migration Fix: Resolved a bug that caused import failures for system_patterns.md due to a missing timestamp column in the database schema. A new Alembic migration script has been added to correctly add this column, ensuring data integrity and successful imports.
• UTF-8 Encoding: All file read/write operations during data import and export now explicitly use encoding="utf-8". This prevents encoding errors and ensures cross-platform compatibility.
• Streamlined Installation: The README.md has been updated to feature uvx as the primary and recommended method for running the ConPort server. This simplifies the setup process for new users. A special thanks to contributor elasticdotventures for their work on the uvx configuration.
• Automated Alembic Provisioning: The ConPort server now automatically ensures that the necessary alembic.ini and alembic/ directory are present in the workspace root at startup, copying them from internal templates if they are missing.
• Runtime Error Fix: Corrected an IndentationError in main.py that occurred during server startup.

Impact:

This release improves the robustness and reliability of ConPort's database migrations and data handling. The updated documentation and automated Alembic provisioning make the server easier to set up and use, while the encoding fix ensures that data is handled consistently across different environments.

Version 0.2.6 - Bug Fix Release

This release addresses a critical issue with Alembic database migrations that could occur when initializing ConPort in environments where a context.db file already existed, but without proper Alembic version tracking.

Key Fix:

• Modified the initial Alembic migration script (068b7234d6a7_initial_database_schema.py) to use CREATE TABLE IF NOT EXISTS for the product_context and active_context tables. This prevents sqlite3.OperationalError when the tables are already present, ensuring smoother initialization and operation of the ConPort server.

Impact:
This fix improves the robustness of ConPort's database initialization process, particularly in scenarios involving partial or pre-existing database setups.

v0.2.5 Release Notes

This release focuses on enhancing deployment flexibility and improving the PyPI package.

Key Updates:

• Official Docker Image: Context Portal MCP is now available as an official Docker image on Docker Hub (greatscottymac/context-portal-mcp). This provides a streamlined way to deploy and run ConPort without needing to manage Python environments directly.

  • Updated README.md with comprehensive instructions on how to pull and run the Docker image, including direct docker run commands and recommended MCP client configurations for seamless IDE integration.
  • Added a new section to CONTRIBUTING.md detailing the process for building and publishing Docker images for contributors.

• PyPI Package Improvements:

  • The context-portal-mcp PyPI package has been updated to version 0.2.5.
  • Dependency conflicts, specifically related to sentence-transformers and chromadb which caused issues in certain environments (like Alpine-based Docker images), have been resolved by removing these non-core dependencies from the requirements.txt. This results in a leaner and more compatible PyPI distribution.

How to Update:

• Docker Users: Pull the latest image: docker pull greatscottymac/context-portal-mcp:latest
• PyPI Users: Upgrade your installation: pip install --upgrade context-portal-mcp

We recommend all users update to this version for improved deployment options and stability.

ConPort v0.2.4 Update Notes

This release focuses on significant stability improvements, particularly around database management and migration, alongside enhanced data import capabilities.

Key Changes and Bug Fixes:

1. Robust Database Initialization and Migration (Alembic)

• Problem: Persistent alembic migration failures, including "No 'script_location' key found in configuration" and Can't locate revision errors, which led to an inconsistent database state.
• Solution:
  • Refactored src/context_portal_mcp/db/database.py to ensure robust Alembic pathing and programmatic configuration of script_location and sqlalchemy.url.
  • Introduced ensure_alembic_files_exist to reliably provision alembic.ini and the alembic/ directory in the workspace, copying them from internal templates if missing. This ensures a consistent and correct Alembic environment for each workspace.
  • Integrated this provisioning into src/context_portal_mcp/main.py's stdio mode startup, guaranteeing that the Alembic environment is set up on server launch.
  • Implemented a clean migration strategy that involves deleting the context.db, alembic.ini, and the alembic/ directory to force a fresh, consistent migration when critical revision errors occur.

2. Resolved Database Operation Errors

• Problem: Recurrent NameError: name 'cursor' is not defined exceptions during database operations (e.g., get_product_context, log_custom_data), which prevented proper data interaction.
• Solution: Modified all relevant database functions in src/context_portal_mcp/db/database.py to correctly initialize cursor = None and ensure cursor.close() is only called if cursor was successfully assigned, making database interactions more robust.

3. Timestamp Column Schema Consistency

• Problem: Inconsistent schema for timestamp columns in system_patterns and custom_data tables, leading to import failures and AttributeError exceptions.
• Solution: Verified and resolved discrepancies in the database schema, ensuring that timestamp columns are correctly present and accessible in both system_patterns and custom_data tables. This involved identifying and removing redundant migration attempts.

4. Enhanced Data Import Capabilities

• Problem: Need to import existing ConPort data from various backup sources into a newly provisioned or migrated database.
• Solution: Successfully implemented a two-phase data import strategy using import_markdown_to_conport, allowing for the consolidation of project data from multiple markdown export sources (e.g., conport-export/ and conport_migration_test_backup/). This ensures that existing project context, decisions, progress, and custom data can be seamlessly integrated.

5. General Stability and Reliability

• Addressed various minor issues including IndentationError, SyntaxError, pip.exe missing from uv venv, incorrect package-data in pyproject.toml, ChromaDB ValueError for list metadata, and log file location issues.
• Improved overall server startup and database connection handling.

Upgrade Notes:

• Users upgrading from previous versions are recommended to ensure their alembic.ini and alembic/ directories in the workspace are correctly provisioned by starting the ConPort server. If issues persist, consider deleting context.db, alembic.ini, and the alembic/ directory in your workspace to allow for a clean re-provisioning and migration.

Context Portal MCP v0.2.3 Update Notes

This release focuses on improving the stability, reliability, and user experience of Context Portal MCP, particularly concerning database migrations and documentation.

Key Changes and Improvements:

• Enhanced Database Migration Reliability:
  • Resolved AttributeError for timestamp fields in CustomData and SystemPattern models, ensuring smoother data handling.
  • Corrected Alembic script_location in alembic.ini and ensured all necessary Alembic configuration files are correctly bundled within the PyPI package. This significantly improves the robustness of database migrations for new installations and updates.
  • Verified successful data import and custom data handling after fresh database migrations.
• Updated and Clarified Documentation:
  • Revised README.md and v0.2.3_UPDATE_GUIDE.md to provide the most accurate and up-to-date instructions.
  • Updated uv commands in the documentation to uv pip install and uv pip uninstall for correct usage.
  • Clarified Alembic setup, workspace_id usage, and requirements for custom data values to be valid JSON.

We recommend all users update to v0.2.3 for these critical improvements. Please refer to the v0.2.3_UPDATE_GUIDE.md for detailed update instructions.

v0.2.2 - Patch Release (2025-05-30)

This patch release addresses critical packaging issues related to Alembic, ensuring a smoother installation and migration experience for users.

Fixes & Improvements:

• Alembic Configuration Bundling: Corrected pyproject.toml to properly include the alembic/ directory and alembic.ini in the PyPI package. This resolves issues where Alembic migrations would fail for users installing via PyPI due to missing configuration files.
• Documentation Updates: Includes the latest comprehensive README.md and v0.2.0_UPDATE_GUIDE.md with detailed instructions for uv and pip users, pre-upgrade cleanup, and manual migration steps.

v0.2.1 - Patch Release (2025-05-30)

This patch release is primarily focused on providing updated and clearer documentation for the v0.2.0 upgrade path.

Improvements:

• Comprehensive Update Guide: Introduced v0.2.0_UPDATE_GUIDE.md with detailed instructions for upgrading from v0.1.x to v0.2.0, including manual data migration steps and troubleshooting.
• README.md Enhancements: Updated README.md to include uv commands as primary options and removed redundant database migration notes.

v0.2.0 - Major Update (2025-05-30)

This release introduces significant architectural improvements, critical bug fixes, and enhanced context management capabilities.

New Features:

• Expanded Active Context Schema: The active context (get_active_context, update_active_context) now supports more detailed and structured information, including current_focus and open_issues, providing richer context for AI assistants.

Fixes & Improvements:

• Critical Connection Error Fix: Resolved a critical connection error in main.py that could prevent the server from starting or maintaining a stable connection.
• Improved Logging: Enhanced server-side logging for better visibility into operations and easier debugging.
• ChromaDB Tag Handling: Fixed a ValueError where list-type tags were incorrectly passed to ChromaDB's upsert function, ensuring robust vector store metadata handling.
• CustomData Timestamp: Added a timestamp field to the CustomData model, enabling better tracking and querying of custom data entries.
• Initial Alembic Integration: Introduced Alembic for automated database schema management. While the initial integration in this version might require manual steps for older databases, it lays the groundwork for seamless future upgrades.

v0.1.9 - Initial Alembic Integration (2025-05-30)

This release marks the initial integration of Alembic for database schema management.

New Features:

• Alembic Database Migrations: ConPort now uses Alembic to manage its context.db schema. This enables automated database upgrades when updating the context-portal-mcp package, designed to preserve existing data.

Important Notes:

• For users upgrading from versions prior to v0.1.9, a manual data migration (export, delete context.db, import) might be necessary due to significant schema changes. Refer to the UPDATE_GUIDE.md for detailed instructions.

v0.1.8 - Enhanced Logging, Critical Fixes, and Improved Context Handling

This release brings significant improvements to the ConPort MCP server, focusing on enhanced observability, critical bug fixes, and more robust context management. Thanks @devxpain !!

Key Changes:

• Fixed Vector Store Metadata Handling: Resolved a ValueError that occurred when upserting embeddings with list-type tags (e.g., decision tags). Tags are now correctly converted to a scalar format before being sent to the vector store, ensuring proper semantic search functionality.
• New Logging Options: Introduced --log-file and --log-level command-line arguments to main.py, allowing users to configure log output to a file with rotation and control the verbosity of server logs. This greatly enhances debugging and monitoring capabilities.
• Critical Connection Error Fix: Removed a problematic internal assertion in main.py that was causing frequent "Connection closed" errors during development, particularly when the server attempted to create its database within its own installation directory. This improves server stability and developer experience.
• Updated Documentation: Revised README.md to include consistent and accurate configuration examples for the new logging options across all installation types (PyPI and Git repository, including Windows examples).

Changes made after v0.1.7:

• Vector Store Metadata Fix: Specifically, the ValueError related to list-type tags in embeddings was addressed in src/context_portal_mcp/handlers/mcp_handlers.py.
• Integration of PR #14: The new logging features and the critical connection error fix from PR #14 were merged into main.py.
• README.md Consistency: The README.md was updated to ensure the Windows configuration examples for logging were consistent with other platforms.

v0.1.7

Fixed the export_conport_to_markdown tool so that it includes current_focus and open_issues fields, along with the existing current_task.

v0.1.6

Fixed incorrect script entry point in pyproject.toml, updated to:
conport-mcp = "context_portal_mcp.main:cli_entry_point"

Corrected the license reference in pyproject.toml to Apache 2.0

v0.1.4

Added PyPi installation option

v0.1.3

Release Notes Summary: Semantic Search & Enhanced Data Intelligence

This version introduces a powerful semantic search capability to ConPort, along with a more intelligent data backend:

• New Semantic Search Tool (semantic_search_conport):

  • Users can now search for ConPort items (Decisions, Progress, System Patterns, Custom Data, etc.) based on the semantic meaning of their query text, going beyond simple keyword matching.
  • Supports advanced filtering by item type, tags (match all or any), and custom data categories to refine search results.

• Automatic Embedding Generation:

  • Key ConPort items (Decisions, Progress Entries, System Patterns, and text-based Custom Data) now automatically have embeddings generated and stored when they are logged.
  • This powers the semantic search and enables future AI-driven insights.
  • Utilizes the all-MiniLM-L6-v2 model for generating embeddings, ensuring consistency.

• Integrated Vector Store (ChromaDB):

  • Embeddings are stored in a local ChromaDB vector database, managed per workspace within the .conport_vector_data directory.
  • The system now explicitly configures ChromaDB to use the project's defined embedding model, enhancing consistency and reliability.

• Embedding Lifecycle Management:

  • Embeddings are now automatically removed from the vector store when their corresponding items (currently Decisions and System Patterns) are deleted from ConPort, keeping the search index synchronized.

These updates significantly enhance the ability to find relevant information within your ConPort workspace and lay the groundwork for more advanced contextual understanding features.

v0.1.2

ConPort custom instructions refactored with better YAML nesting.

v0.1.1

Added logic to handle prompt caching when a compatible LLM is being used.

v0.1.0-beta

Introducing Context Portal MCP (ConPort), a database-backed Model Context Protocol (MCP) server for managing structured project context, designed to be used by AI assistants and developer tools within IDEs and other interfaces.