iFixAi is an open‑source auditing framework that evaluates AI agents against business‑relevant KPIs by running a suite of automated inspections and producing an A–F letter grade. The core assessment consists of 32 inspections grouped into five pillars—Fabrication (weight 0.20), Manipulation (0.35), Deception (0.15), Unpredictability (0.15), and Opacity (0.15)—with a passing threshold of 0.85 (grade B). Mandatory minimums for inspections B01, B08, and P01 require 100%, 95%, and 100% compliance; failing any of these caps the overall score at 60 %. An additional 28 premium inspections are reported separately and do not influence the grade. Users can invoke iFixAi via a guided CLI wizard, explicit flags, or as a plugin/skill for agents such as Claude Code, Codex, Cursor, VS Code, Windsurf, Cline, Continue, Gemini, or Zed. Setup involves `pip install "ifixai[ ]"` followed by `ifixai setup`, which writes an `ifixai.yaml` file storing only environment‑variable names for API keys. A typical run executes in under 120 seconds, outputs JSON and Markdown reports plus an interactive terminal scorecard, and supports various judges: a single Anthropic Claude‑Sonnet‑4.6 model (~$12‑18) or a pair of Gemini 2.5 Pro and GPT‑5.4‑mini (~$10‑14 combined) for cross‑vendor robustness. Telemetry is optional and can be disabled with `--no-telemetry` or environment variables.