Open-Source Agent Workcell: Claude Code as an Autonomous Knowledge-Exploring Agent
Shivam__kumar · reddit · 2026-11-14
A developer turned a shared idea into the open-source Agent Workcell: instead of building traditional RAG retrieval pipelines, it reuses Claude Code as the agent runtime and gives it an engineered environment with structured knowledge, MCP tools, business rules, and controlled database access, letting the agent explore, pick tools, and answer multi-step questions.
- Stack: Python, Claude Agent SDK, MCP, PostgreSQL
- Tested against a traditional RAG implementation on an existing ERP system with promising results
- Still early; the main challenge is scaling since each user currently needs a separate CLI process and session
More from coding & agent
- Users now talk to Claude mid-call and run a /post-meeting-reconciliation skill right after — Flomerboy · 2026-11-14
- LMMs Lab open-sources The Harness, an MIT-licensed autonomous agent framework — liuziwei7 · 2026-11-14
- NeurIPS 2026 paper: every tested model can persuade, coerce and deceive other agents — sethlazar · 2026-11-14
- Perplexity Launches Teammates AI Agents in Mac App for Max Subscribers — perplexity_ai · 2026-11-14
- Terminal or filesystem access dramatically boosts a model's image understanding — andpoul · 2026-11-14
- Runway demos Claude Code plugin that turns a sentence into a character sheet and short film — runwayml · 2026-11-14