@mind-as-context/mcp-server

Record knowledge context from the current work session. Use this to capture: - Implementation details (commands executed, code changes made) - Planning decisions (design choices, architecture decisions) - Thoughts (ideas, assumptions, reasoning behind decisions) - Intents (user goals, requirements, acceptance criteria) - Decisions (technology choices, tradeoffs, alternatives considered) This tool helps maintain project context across sessions.

Query existing knowledge from the context store. Use this to: - Find related context before starting work - Check existing implementations - Understand project decisions and reasoning - Search for specific topics or code references

Query and search recorded decisions from the knowledge base. Use this to: - Find past technology and architecture decisions - Understand why certain choices were made - Discover alternatives that were considered - Check for related decisions before making new ones Returns decisions with their rationale, alternatives considered, and tradeoffs.

Check for outdated knowledge that may need updates. Returns documents that might be stale due to: - Recent code changes that invalidate documentation - Time elapsed since last review - Dependency changes (outdated dependencies) Use this before and after making changes to identify what documentation needs updating.

Update relationships between knowledge nodes or change their status. Actions: - link: Create a relationship between two knowledge nodes - unlink: Remove a relationship between two knowledge nodes - mark_current: Mark a knowledge node as current (up-to-date) - mark_outdated: Mark a knowledge node as outdated - archive: Archive a knowledge node (no longer relevant) Relation types: - references: Source references target as supporting information - implements: Source implements the design described in target - depends_on: Source depends on target being accurate (propagates outdated status) - supersedes: Source replaces target - related_to: Bidirectional relationship - derived_from: Source was created based on target - contradicts: Source conflicts with target (content mismatch) - fulfills: Source fulfills the intent in target - extends: Source extends the intent in target with additional scope - revises: Source revises or updates the intent in target

Analyze an AI agent work session transcript and extract structured knowledge. This tool automatically: - Extracts decisions made during the session - Identifies implementations and code changes - Captures insights, ideas, and concerns - Suggests relevant tags and connections Use this at the end of a work session to capture important context.

Find related knowledge based on semantic similarity. Use this tool to: - Discover relevant past decisions before making new ones - Find related implementations and planning documents - Understand the context around a specific topic - Get suggestions for linking knowledge nodes Returns related knowledge nodes ranked by relevance.

Check consistency between decisions in the knowledge base. Use this tool to: - Find conflicting decisions in the same scope - Detect overlapping or redundant decisions - Identify dependency issues between decisions - Discover outdated decisions that need review Returns issues found and suggestions for resolution.

Trace the origin path from any knowledge node back to its intent/decision roots. Use this tool to: - Understand why a particular implementation exists - Find the original intent behind a piece of code - Discover the decision chain that led to current state - Audit the reasoning behind implementations Returns the ancestry chain from the target node back to its origins.

Query implementation documentation for codebase understanding. Use this BEFORE reading actual source code to understand: - Module structure and purpose - Key interfaces and types - API signatures - Code patterns and conventions This returns compressed documentation optimized for token efficiency, reducing context usage by ~90% compared to reading source files directly. Load levels: - index: Project overview only (~200 tokens) - summary: Architecture summary (~800 tokens) - module: Specific module docs (~1500 tokens each) - detail: Full documentation with examples (~3000+ tokens)

Analyze and generate/update implementation documentation. Use this to: - Generate initial impl documentation - Update documentation after code changes - Get codebase statistics Actions: - stats: Get project statistics (files, lines, complexity) - status: Check documentation status (up-to-date vs outdated) - analyze_agent: Run agent-based analysis to generate/update documentation