Giskard
EvaluationApache-2.0
Giskard tests LLM agents with checks, LLM judges and a scan for vulnerabilities and quality problems. Giskard publishes it as an open-source Python library, so the tests run inside the developer's own code and pipelines.
The README splits the project into installable parts: checks together with agents, llm and core parts, a scan extra, plus provider extras for model judges. The GitHub repository record is now named giskard-oss.

Pinout
- MCP server: not checked yet
- HTTP API: not checked yet
- command line: not checked yet
- Python SDK: supported, verified
- TypeScript SDK: not checked yet
- OpenAPI description: not checked yet
| Interface | Status | Source we read | Read on |
|---|---|---|---|
| MCP server | Not checked yet | — | — |
| HTTP API | Not checked yet | — | — |
| Command line | Not checked yet | — | — |
| Python SDK | Supported, verified | README, install (pip install giskard) | |
| TypeScript SDK | Not checked yet | — | — |
| OpenAPI description | Not checked yet | — | — |
Facts
| License | Apache-2.0. GitHub API, repository license field, read . |
|---|---|
| Main language | Python (as detected by GitHub). GitHub API, repository record, read . |
| Publisher | Giskard-AI (GitHub organization). GitHub API, repository record, read . |
| Latest GitHub release | giskard-agents/v1.0.6, published 2026-10-25. GitHub API, latest release, read . |
| Install extras | giskard, giskard[scan], giskard[openai], giskard-checks. README in the repository, read . |
| License (README badge) | Apache 2.0. README in the repository, read . |
| Offerings | Giskard Hub (UI and SDK) and Giskard Open Source (Python library). Giskard documentation, overview, read . |
| Scan | Not installed by default; added with the scan extra (pip install "giskard[scan]"). Giskard documentation, overview, read . |
| Hub on-premise | Offered for mission-critical workloads in the public sector, per the FAQ of the home page. Giskard home page, FAQ, read . |
| Compliance stated by the vendor | GDPR, SOC 2 Type II and HIPAA, as stated on the home page. Giskard home page, Compliance & Security, read . |
| Last commit | . GitHub API, latest commit on the default branch of Giskard-AI/giskard-oss, read . |
Limits
- Judge-based checks such as groundedness call a model provider, so a provider extra and its API key are needed, per the README.
Strengths
- Checks, scan and provider support are separate extras, so only the parts in use get installed.
Alternatives to Giskard
All evaluation tools for agents
Open-core platform to trace, test and evaluate agents and coding assistants.

Framework for evaluating, testing and monitoring ML and LLM systems.

Framework for LLM evaluations from the UK AI Security Institute.

Command line tool and library to evaluate and red-team LLM apps and agents.
Last check: . Method: public repository and docs read automatically, page and day written down. Next check due within 90 days. How we check
Is this your tool? Write to hello@agent-tools.directory from an address on its domain to correct a pin or a fact.