ManzoliW/effect-harness-agent

Effect Harness Agent is an agent-driven engineering framework for TypeScript/Effect projects — an LLM-native, machine-to-machine protocol (not a human-facing framework). It turns natural-language intent into structured, executable tasks for autonomous LLM agents while enforcing engineering discipline. Who uses it: teams or systems that integrate LLM agents to design, implement, test, and maintain TypeScript/Effect code with predictable governance and cost. Key features: - Intent Gateway routing and execution profiles (@patch/@standard/@learn) - 6‑phase Lifecycle State Machine with human-in-the-loop approval and validation gates - Knowledge Graph (LLM Wiki) and budgeted navigation to avoid context bloat - Contract-first OpenSpec and explore_report outputs before coding - Skills Matrix (specialized agent capabilities) and self-correcting guard hooks - Token-cost model and shortcuts to control cost and prevent runaway searches

AI-related keywords found in this repo's description, topics, or README summary — grouped by category. Each badge links to the corresponding ranking detail page.