The Context Engine MCP Server is a local-first, agent-agnostic platform for workspace indexing, semantic search, AI-assisted planning, code review, memory management, and security validation.

Workspace Indexing & Retrieval

• Index 50+ file types for semantic search (index_workspace, semantic_search, codebase_retrieval)

• Retrieve full/partial file contents (get_file) and token-aware context for AI prompts (get_context_for_prompt)

• Enhance simple prompts into detailed, codebase-aware prompts (enhance_prompt)

• Manage index health: check status, reindex, or clear (index_status, reindex_workspace, clear_index)

Memory Management

• Persist preferences, architecture decisions, and project facts across sessions (add_memory, list_memories)

Planning & Execution

• Generate AI-powered implementation plans with dependency graphs, risk assessments, and Mermaid diagrams (create_plan, refine_plan, visualize_plan)

• Execute plan steps with AI-generated code changes, preview or apply (execute_plan)

• Save, load, list, and delete plans (save_plan, load_plan, list_plans, delete_plan)

• Track step-by-step progress and version history; roll back plan versions (start_step, complete_step, fail_step, view_progress, view_history, compare_plan_versions, rollback_plan)

• Gate execution with human approval workflows (request_approval, respond_approval)

Code Review

• AI-powered review of diffs or git changes with structured findings (P0–P3 priority, confidence scores) (review_changes, review_diff, review_git_diff, review_auto)

• Reactive PR review with parallel execution, session management, commit-aware caching, and telemetry (reactive_review_pr, get_review_status, pause_review, resume_review, get_review_telemetry)

• Deterministic invariant checks against YAML-defined rules (check_invariants) and static analysis via TypeScript/Semgrep (run_static_analysis)

Security & Validation

• Detect and mask 15+ secret types before sending content to an LLM (scrub_secrets)

• Multi-tier content validation: bracket balancing, JSON structure, TODO detection, hardcoded URLs, and automatic secret scrubbing (validate_content)

Tool Discovery

• Inspect all available server tools and capabilities (tool_manifest)

Provides comprehensive tools for automatic retrieval and review of code changes, including staged, unstaged, branch, and commit-level diffs.

Enables the visualization of structured execution plans by generating and displaying Mermaid-formatted diagrams.

Integrates with local static analysis tools like tsc to provide deterministic feedback and analysis during the code review process.

Supports deterministic invariant checking using YAML-based rule files to enforce project-specific constraints during reviews.

1. Click on "Install Server".

2. Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.

3. In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Context Engine MCP Server find where the user authentication logic is implemented"

That's it! The server will respond to your query, and you can continue using it as needed.

Here is a step-by-step guide with screenshots.