Show HN: A lightweight app to let LLM work for oncall
NeatContext is a local-first desktop workspace that integrates domain-specific knowledge and internal systems with Large Language Models to improve operational accuracy. The tool addresses the gap between generic AI responses and specific organizational context by allowing users to define Markdown-based profiles for different teams or services. It supports extensibility via Model Context Protocol (MCP) and connects to various LLM providers (OpenAI, Anthropic, Ollama) while keeping data local for
Analysis
TL;DR
- NeatContext is a local-first desktop workspace that integrates domain-specific knowledge and internal systems with Large Language Models to improve operational accuracy.
- The tool addresses the gap between generic AI responses and specific organizational context by allowing users to define Markdown-based profiles for different teams or services.
- It supports extensibility via Model Context Protocol (MCP) and connects to various LLM providers (OpenAI, Anthropic, Ollama) while keeping data local for security and control.
- Demonstrated use cases show significant improvements in incident resolution, such as correctly identifying root causes in infrastructure outages versus providing generic checklists.
- The architecture emphasizes lightweight setup, read-only tool access, and version-controlled knowledge bases to ensure trust and reproducibility in AI-assisted operations.
Why It Matters
This solution highlights the critical industry shift toward grounding LLMs in proprietary, real-time operational data rather than relying solely on pre-trained general knowledge. For AI practitioners and DevOps engineers, it demonstrates how local-first architectures can mitigate data privacy risks while enhancing the precision of automated decision-making in high-stakes environments.
Technical Details
- Local-First Architecture: All profiles, indexes, conversations, and tool logs are stored locally on the user's machine, ensuring data sovereignty and reducing latency associated with cloud-based retrieval.
- Markdown-Based Knowledge Profiles: Users create structured Markdown files to define domain expertise, service ownership, and runbooks, which are easily version-controlled and editable.
- MCP Integration: Supports the Model Context Protocol to connect LLMs with read-only internal tools and systems, enabling verified evidence gathering during analysis.
- Model Agnostic: Compatible with any OpenAI-compatible API, Anthropic, Ollama, or private gateways, allowing organizations to leverage existing model investments.
- Contextual Reasoning: The system provides "grounded" answers by citing specific document lines (e.g.,
postgres-connection-pool.md:38-42) and verifying actions against historical incident data.
Industry Insight
- Operational AI Adoption: Organizations should prioritize tools that bridge the gap between static documentation and dynamic AI interaction, especially for SRE and DevOps workflows where context is king.
- Security by Design: The local-first approach offers a viable path for regulated industries to adopt LLMs without exposing sensitive internal data to third-party clouds, balancing innovation with compliance.
- Standardization of Context: The use of Markdown for knowledge profiles suggests a trend toward standardized, human-readable formats for AI context management, making it easier for non-engineers to contribute to and maintain AI-enhanced workflows.
Disclaimer: The above content is generated by AI and is for reference only.