Directory / Giskard
Giskard
Verified
Open testing and red-teaming for AI applications.
Giskard is an open-source platform for testing ML and LLM applications: functional tests, scanning, RAG evaluation. It sits in verification when the factory’s output is an AI feature, not only a patch to a CRUD service.
- Model independence
- Multi-model
- Deployment
- Cloud, Self-hosted
- Open source
- Yes
- License
- See repository (open-core)
- Pricing model
- Mixed
- Added / updated
- 19 Aug 2026 / 19 Aug 2026
www.giskard.ai · Docs · Source
Editorial take
Giskard is for when the software under test *is* the model. Most coding factories still need Semgrep first. Open-core: read the license. Not a PR comment bot.
Strengths
- AI-app testing beyond a single prompt file.
- Self-host path.
- Scan/red-team features documented.
Limitations
- Wrong tool for a generic backend diff.
- Open-core split.
- Not a sandbox.