Validate and retry LLM outputs for ruby_llm. Describe the JSON response you expect, fall back to a stronger model when the cheaper one fails the rules, and gate CI on regressions — all as one contract object per step.
For AI enthusiasts who try every new tool and watch their skills fragment. One source of truth for AI agent skills (SKILL.md) + cross-tool sync toolkit: MCP server, shared instructions, and one-command wiring for Claude Code, Codex, Gemini, Kimi, Cursor, Kilocode, OpenCode.
Pin feedback to the live UI. Hand it to Claude. An embeddable visual-feedback SDK — inspect an element, pin a comment, capture a screenshot; comments re-anchor across redeploys and reach Claude Code via MCP.
Open-source data protection framework for AI agent workflows in n8n. Detects, tokenizes, and reversibly anonymizes sensitive data before it reaches your models.
AI vision GUI automation for browsers, mobile apps, desktops, and games — no DOM, no selectors. Standalone or on top of Playwright / Selenium / Appium / pytest.
OpenCode plugin: use AWS Kiro (CodeWhisperer) Claude & open-weight models via your Kiro subscription. Provider id kiro-auth. Multi-account rotation, usage display, Sonnet 5 / Opus 4.x.