twilliamsliu/qa-agent-framework

A platform-layer framework of conventions for running multiple AI Agents in QA workflows — not a tool or MCP, but a set of structures and guardrails that sit above your LLM/IDE. It’s intended for solo QA engineers, small teams, or anyone scaling a PoC beyond single-agent prompt hacks. Key goals: prevent agent drift, conflicting edits, repeated errors, and secret leakage. Core elements (six concerns): Agent Topology (Hub-and-Spoke orchestrator + per-project agents); Knowledge Architecture (three-tier memory: global/project/agent); Governance (iron laws, heuristic rules, self-learning rules captured at startup); Sync & GitOps (three-tier repo layout, pre-commit, push/sync helpers); Secret & Cloud (L1 shareable → L3 local-only); Tool Capability (swappable integrations). Status: early, N=1 verification; vendor-agnostic by design. Use only when coordination problems warrant it.

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.