NoWorries — Open-source AI desktop assistant
A solo-built desktop-grade autonomous Agent: give it a plain-language instruction, and it plans, picks tools, and executes across multiple steps on its own.
Problem
Repetitive desktop work — sorting files, driving Office apps, looking things up — has always lacked one thing: a local Agent that can plan and execute on its own, and that you can actually trust with real permissions. "Trusting it with permissions" means solving three hard problems first: memory, safety boundaries, and rollback.
Approach
Built on top of OpenWork (Electron + TypeScript + Python). Three core designs: (1) a three-tier memory architecture — instant / episodic / core — backed by vector embeddings and semantic search, with incremental summarization, time decay, and emotional tagging, to carry context across sessions and personalize over time; (2) a safety execution sandbox — allowlisted workspace isolation, high-risk command interception, sensitive-path protection, automatic backups before any file change, fully auditable end-to-end logs, and one-click rollback; (3) a directory-convention skill plugin system that auto-discovers and registers tools at runtime, invokes them dynamically via Function Calling, automates Excel/Word/PPT, and lets plugins be developed independently and hot-reloaded.
Results
Open-source + live website
Form
Three tiers (instant / episodic / core)
Memory architecture
Sandbox + backup-and-rollback + end-to-end logs
Safety design
AI's role in this project
The project itself is Agent engineering: memory, safety boundaries, and tool calling are all designed and implemented by me — the most direct proof of actually knowing how to build with LLMs.
Why I built it
Most AI assistants stop at "chat." I wanted one that just does the work: say "tidy up my Downloads folder by file type," and it plans the steps, calls the tools, and finishes the job. The hard part isn't calling the LLM — it's three engineering problems: Does it remember? Is it safe? Can a mistake be undone?
Three core designs
1. Three-tier memory architecture. Instant memory (the current session), episodic memory (events across sessions), and core memory (long-term user preferences), all built on vector embeddings + semantic search. Paired with incremental summarization, time decay, and emotional tagging — so over long-running use, memory stays compact, stays current, and keeps its priorities straight.
2. Safety execution sandbox. This is the boundary control for an Agent acting on its own: allowlisted workspace isolation, high-risk command interception, sensitive-path protection; automatic backups before any file change, fully auditable operation logs, and one-click rollback. "Daring to let an Agent touch your files" rests entirely on this layer.
3. Pluggable tool system. Skill plugins load by directory convention: drop a tool module into the directory and the runtime auto-discovers and registers it, while the Agent selects and calls it dynamically via Function Calling. Office automation — Excel, Word, PPT — all lives as plugins that can be developed independently and hot-reloaded.
Links
- Source: github.com/hlbbbbbbb/No-worries
- Website: meetnoworries.com