Giskard-OSS is an open-source Python library for testing and evaluating agentic systems, with modules for checks, scanning, and RAG evaluation. This review lists installed packages, usage snippets, and release notes from the README and releases data.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
Giskard-OSS is an open-source Python library for testing and evaluating agentic systems. The v3 architecture is a modular set of focused packages — each carrying only the dependencies it needs — built from scratch to wrap anything: an LLM, a black-box agent, or a multi-step pipeline.
How it works
- The repository comprises multiple packages including giskard-checks (testing & evaluation with scenario API), giskard-scan (vulnerability scanning for agents), and planned giskard-rag (RAG evaluation).
- Giskard Checks uses Scenario, Groundedness, and LLMJudge styles to validate non-deterministic outputs and groundedness against retrieved context.
- Giskard Scan provides red-teaming capabilities to generate adversarial inputs and run prompt-injection probes across OWASP LLM Top-10 categories.
Getting started
Install options shown in the README:
pip install giskard # checks (+ agents, llm, core)
pip install "giskard[scan]" # + vulnerability / quality scan
pip install "giskard[openai]" # provider SDK for LLM judges / generators
Requires Python 3.12+.
Telemetry is optional and can be disabled before importing Giskard:
export DO_NOT_TRACK=1
or
export GISKARD_TELEMETRY_DISABLED=1
Recent releases
Latest releases include:
- giskard-scan/v1.0.0b3 (2026-07-13) with fixes for CI checks and tag copying in garak probes
- giskard-llm/v1.0.0b5 (2026-07-13) with full changelog link
- giskard-core/v1.0.1b5 (2026-07-13) with new scenarios generator and scenario tag grouping
- giskard-checks/v1.0.2b5 (2026-07-13) with CI-related fixes
- giskard-agents/v1.0.2b5 (2026-07-13) with full changelog link
Traction
Stars: 5735, Forks: 513, Open issues: 76 (as per repository stats).
Behind the repo
Not provided in the materials.
Caveats
License: Apache-2.0. Created: 2022-03-06. Last push: 2026-08-04. Language: Python. Topic tags include agent-evaluation, ai-security, llm-evaluation, ml-testing, red-team-tools, trustworthy-ai, etc.
Usage examples include:
- Giskard Checks quickstart:
import asyncio
from giskard.checks import Scenario, Groundedness
- Giskard Scan quickstart:
import asyncio
from giskard.scan import vulnerability_scan
Additional installation guidance and docs are linked in the README for deeper usage.