skills.rest
infektyd
infektydCOMMUNITY
Explore all skills by this author →
Website𝕏@infektydGitHub

minni-engine

Document Minni engine v3.1/v3.2 core pipeline modifications with interface signatures and backend swap patterns.

2|Updated Mar 31, 2026
Search Tags:#retrieval#sqlite#extraction#reranking#faiss#memory system#minni
One-click install
npx skills add https://github.com/infektyd/minni --skill minni-engine
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: minni-engine
Source: https://github.com/infektyd/minni/tree/main/plugins/minni/skills/minni-engine
Command: npx skills add https://github.com/infektyd/minni --skill minni-engine

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Developers modifying Minni's core memory system pipelines lack a centralized, authoritative reference, leading to broken interfaces, incompatible changes, and avoidable errors when adjusting extraction, reranking, consolidation, or retrieval functionality.

Core Features & Use Cases

  • Critical Interface Documentation: Exact signatures and usage patterns for extraction, retrieval, and writeback modules that must not be altered to maintain downstream compatibility.
  • Backend Swap Patterns: Step-by-step guidance for replacing the reranker or extraction LLM backend without breaking existing scoring, fallback, or prompt parsing logic.
  • Development Safeguards: Testing workflows, common pitfall warnings, and configuration update requirements to prevent daemon downtime or agent compatibility issues.
  • Use Case: A developer adding a custom reranker API can follow the documented V3.2 swap pattern to integrate the new service while preserving RRF fallback scores and candidate dict formatting.

Quick Start

Use the minni-engine skill to modify the Minni retrieval pipeline's reranker backend following the documented V3.2 swap pattern.

Frequently Asked Questions about minni-engine

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I swap the LLM backend in a memory retrieval pipeline without breaking reranking?▼

To swap the LLM backend safely, replace the extraction or reranker service using the documented V3.2 integration pattern, which preserves RRF fallback scores and candidate dict formatting to maintain downstream agent compatibility.

What are the critical interfaces to preserve when modifying memory extraction logic?▼

When modifying memory extraction logic, you must preserve the exact signatures and usage patterns of extraction, retrieval, and writeback modules to prevent breaking downstream daemon operations and agent dependencies.

Does adding a new API endpoint to the Minni memory engine require configuration updates?▼

Adding a new API endpoint requires configuration updates and adherence to specific testing workflows to prevent daemon downtime and ensure compatibility with existing agent operations and downstream dependencies.

How does the reranking fallback mechanism work during a backend swap?▼

The reranking fallback mechanism maintains compatibility by preserving RRF fallback scores and candidate dict formatting during a backend swap, ensuring that scoring logic remains intact even when the primary extraction LLM is replaced.

What testing workflows prevent daemon downtime when adjusting writeback logic?▼

To prevent daemon downtime when adjusting writeback logic, follow the documented development safeguards including specific testing workflows and pitfall warnings that verify interface compatibility and configuration integrity.

When should I not modify the retrieval pipeline interfaces in a memory system?▼

You should not modify retrieval pipeline interfaces when downstream agents or daemon operations depend on existing signatures, as altering extraction, retrieval, or writeback modules risks breaking critical compatibility and causing system failures.

Related Skills

Explore More Software Engineering Skills→
yoloshiiyoloshii
197

clawmem

Provide local embedded search and retrieval for MCP-compatible AI agents.

Community
Advanced
alexfalkowskialexfalkowski
2

change-safety

Review code changes for compatibility risks and security implications before merging.

Community
Advanced
ricardorochericardoroche
2

rag-design-patterns

Standardizes chunking, vector database integration, embedding management, and retrieval reranking for RAG pipelines.

Community
Intermediate
novatech2210-cmdnovatech2210-cmd

V3 Memory Unification

Consolidate SQLite and Markdown memory stores into a unified AgentDB backend with HNSW indexing and embedding-driven retrieval.

Community
Advanced