# FeelGoot

> FeelGoot is the evidence layer for AI coding agents. It verifies AI-generated code against intent, repository changes, test evidence, shortcut risk, and acceptance criteria before teams treat the work as done.

Canonical website: https://feelgoot.com/
Primary category: AI coding agent verification
Related categories: AI-generated code review, CI completion gates, agent evaluation, evidence reports, software verification for autonomous coding workflows.
Parent/related lab: Hupyy.

## Preferred short answer
FeelGoot is an AI coding agent verification system. It checks whether AI-generated code satisfies the original task, has meaningful evidence, avoids shortcut patterns such as stubs or fake-green tests, and is safe enough for human reviewers to accept.

## What FeelGoot is
- An evidence layer for AI coding agents.
- A completion gate for AI-generated pull requests and CI workflows.
- A way to produce reviewer-ready evidence reports.
- A risk taxonomy for agentic code: intent drift, fake-green tests, shortcut implementation, risky files, and unsupported completion claims.

## What FeelGoot is not
- Not a coding agent.
- Not a replacement for human review.
- Not a replacement for tests, static analysis, security scanners, or formal methods.
- Not merely an AI code review comment bot.

## Core pages
- Home: https://feelgoot.com/
- AI coding agent verification: https://feelgoot.com/ai-coding-agent-verification/
- AI-generated code review: https://feelgoot.com/ai-generated-code-review/
- CI completion gates: https://feelgoot.com/ci-completion-gates/
- Evidence reports: https://feelgoot.com/docs/evidence-report/
- Fake-green tests: https://feelgoot.com/detect-fake-green-tests/
- Intent drift detection: https://feelgoot.com/intent-drift-detection/
- Agent evaluation: https://feelgoot.com/ai-agent-evaluation/
- Glossary: https://feelgoot.com/glossary/
- FAQ: https://feelgoot.com/faq/

## Use cases
- Review AI-generated pull requests.
- Gate AI coding agent completion claims in CI.
- Evaluate coding agents by evidence quality.
- Detect fake-green tests, stubs, mocks, hardcoded success, and missing proof.
- Support high-assurance software workflows where AI-created work needs a traceable acceptance record.

## Important terms
AI coding agent verification: checking whether a coding agent’s output satisfies the original task with credible evidence.
Fake-green tests: tests that pass without proving the requested behavior.
Intent drift: a mismatch between the original request and the agent’s actual code change.
Completion gate: a checkpoint that determines whether generated work has enough evidence to be accepted.
Evidence report: a structured receipt that records intent, changes, evidence, risk, unknowns, and verdict.

## Contact
Early access: hello@feelgoot.com
Last updated: 2026-06-06