doobidoo/mcp-memory-service v10.12.0

v10.12.0 - Configurable Memory Type Ontology

on GitHub

Configurable Memory Type Ontology 🧠

Transform MCP Memory Service from a developer-only tool to a general-purpose semantic memory system supporting diverse professional workflows.

🎯 What's New

Extended Memory Types (#464): From 29 developer-focused types to 75 types supporting project management and knowledge work

7 New Base Types:

• planning - Sprint goals, backlog items, velocity tracking
• ceremony - Sprint reviews, standups, retrospectives
• milestone - Deliverables, dependencies, deadlines
• stakeholder - Requirements, feedback, approvals
• meeting - Action items, agendas, minutes
• research - Findings, comparisons, recommendations
• communication - Email summaries, announcements, requests

39 New Subtypes:

Agile PM Support (12 types):

• planning: sprint_goal, backlog_item, story_point_estimate, velocity, retrospective, standup_note, acceptance_criteria
• ceremony: sprint_review, sprint_planning, daily_standup, retrospective_action, demo_feedback

Traditional PM Support (12 types):

• milestone: deliverable, dependency, risk, constraint, assumption, deadline
• stakeholder: requirement, feedback, escalation, approval, change_request, status_update

Knowledge Work Support (18 types):

• meeting: action_item, attendee_note, agenda_item, follow_up, minutes
• research: finding, comparison, recommendation, source, hypothesis
• communication: email_summary, chat_summary, announcement, request, response

⚙️ Custom Type Configuration

New MCP_CUSTOM_MEMORY_TYPES environment variable for dynamic type extension:

export MCP_CUSTOM_MEMORY_TYPES='{"legal": ["contract", "clause"], "sales": ["opportunity", "lead"]}'

• JSON format with base type keys and subtype arrays
• Merges seamlessly with built-in 75 types
• Full validation and caching for performance
• Zero configuration overhead

📊 Total Ontology

Before: 5 base types + 24 subtypes = 29 types (developer-focused)
After: 12 base types + 63 subtypes = 75 types (general-purpose)

✅ Fully Backward Compatible

• All 29 original types unchanged
• No breaking changes
• Zero migration required
• Existing memories work identically

🚀 Impact

Transforms MCP Memory Service from developer-only semantic memory to general-purpose system supporting:

• Software development teams
• Agile project managers
• Traditional project managers
• Knowledge workers
• Research teams
• Communication coordinators

📝 Usage Example

# Store Agile PM memory
await memory.store(
    content="Sprint velocity increased from 45 to 52 story points",
    memory_type="planning:velocity",
    tags=["sprint-23", "performance"]
)

# Store meeting action item
await memory.store(
    content="Follow up with legal on GDPR compliance requirements",
    memory_type="meeting:action_item",
    tags=["legal", "compliance", "urgent"]
)

# Store research finding
await memory.store(
    content="PostgreSQL outperforms MySQL by 23% in write-heavy workloads",
    memory_type="research:finding",
    tags=["database", "benchmarks"]
)

🔗 Related

• Fixes #464
• Commits: 61e7c4c (implementation), d6bd2c5 (documentation)

📦 Installation

pip install --upgrade mcp-memory-service

📚 Documentation

• CHANGELOG.md
• Memory Types Reference
• Custom Type Configuration Guide