Skip to the content
Know your tech before you buy it. Makers: get listed from $49/mo

Giskard

giskard.ai

  • Free plan
  • Open source
  • Linux
  • Web
  • API
  • Self-hosted
8.0 Club score Visit giskard.ai

What Giskard is

Giskard provides tools for finding vulnerabilities in AI agents before they reach production. Its listed assessment areas include prompt injection, data disclosure, inappropriate content, hallucinations, contradictions, and omissions. After an assessment, it produces a structured report with a deploy-or-fix verdict, ranked vulnerabilities, functional scenarios, and a Giskard Label if the agent passes. The Hub is a black-box testing tool for conversational text-to-text agents available through an API endpoint. Giskard describes a remediation process in which its team qualifies findings, discusses severity, prioritizes issues, opens workflow tickets, and reruns tests to check fixes. Enterprise options include on-premise, private cloud, and SaaS deployment. The company states that it offers EU or US data residency and isolation, access controls, audit trails, identity-provider integration, and a zero-training policy, as well as encryption at rest and in transit. It lists GDPR, SOC 2 Type II, and HIPAA compliance. The free plan includes an open-source Python library, local deployment, basic vulnerability scanning and RAG evaluation, community support, and best-effort maintenance.

Who it suits

Giskard suits teams assessing conversational AI agents through an API, particularly those preparing deployments or reviewing security and quality. Enterprise deployment and support options may fit organisations with specific infrastructure or support needs.

Good to know, and what to watch

  • Assessment reports include ranked vulnerabilities and a deploy-or-fix verdict.
  • Enterprise deployment options include on-premise, private cloud, and SaaS.
  • The open-source library runs behavioral tests under pytest.
  • Enterprise plans list dedicated support with SLAs.
  • Hub supports conversational text-to-text agents accessible through an API.
  • The free plan includes best-effort maintenance.
  • No Enterprise price is listed.

Enterprise

—

Professional

—

Free

Free

Open-source

Free

Free plan

Getting started

  1. Open the Giskard website
  2. Use the open-source Python library locally or choose Enterprise
  3. Connect a conversational text-to-text agent through an API endpoint
  4. Run vulnerability or RAG evaluations
  5. Review the assessment report and ranked findings
  6. Rerun tests after remediation

Limits to check

The Free plan provides a basic LLM vulnerability scan using adversarial techniques from 2024 and a basic RAG report using correctness metrics. It includes community support and best-effort maintenance; broader red-teaming, evaluation and dedicated support are listed under Enterprise.

The Club's review

Giskard is an AI-agent testing platform for teams assessing conversational systems before production. It is strongest for teams that can expose an API-accessible text agent and want security findings carried through remediation. Free local tools make it possible to start small, while production use brings a substantially broader, custom-priced Enterprise package.

Overview

Giskard’s Hub performs black-box assessments of conversational text-to-text agents reachable through an API. Its coverage spans prompt injection, data disclosure, inappropriate content, hallucinations, contradictions, and omissions. That makes its scope relevant to teams checking both security and response quality, but the endpoint and conversational format are real fit requirements.

Each assessment returns a structured report with a deploy-or-fix verdict, ranked vulnerabilities, and functional scenarios; agents that pass receive a Giskard Label. Giskard also describes a remediation service: its team qualifies findings, discusses severity, helps prioritize issues, opens tickets in the customer’s workflow, and reruns tests to confirm fixes. This can turn scan results into accountable work, especially for teams that need help managing the response rather than another list of alerts.

Key features

Enterprise red teaming includes more than 50 automated adversarial probes, including multi-turn attacks, with techniques regularly updated. Teams evaluating retrieval-augmented generation can use fine-grained RAG quality metrics and domain-specific evaluation datasets; customizable metrics and human-review interfaces provide room to adapt evaluation to a particular domain. These capabilities are valuable when generic correctness scores are not enough, though they sit in the Enterprise tier rather than the free starting package.

The open-source Python library runs behavioral tests under pytest and includes vulnerability and RAG quality scans, making it useful for teams that want tests in a development workflow. Giskard documents native SDK integrations for OpenAI, Google/Gemini, Anthropic, and Azure, with optional LiteLLM support for other providers. One notable boundary: tabular classification and regression models have no equivalent in the library’s v3.

Enterprise security controls include SSO, role-based access, versioning with audit trails, and scheduled email alerts. Giskard also states that it offers EU or US data residency and isolation, identity-provider integration, and a zero-training policy, alongside encryption at rest and in transit and GDPR, SOC 2 Type II, and HIPAA compliance. These controls may suit sensitive deployments, but organizations should confirm that the deployment and controls match their own requirements.

Pricing

Free — free. The free plan includes the open-source library, local deployment, a basic LLM vulnerability scan using adversarial techniques from 2024, a basic RAG evaluation report using correctness metrics, best-effort maintenance, and community support. It is a credible entry point for local experimentation and baseline checks. The trade-off is a basic scan and report, older stated adversarial techniques, and no dedicated support or Enterprise evaluation and security package.

Enterprise — custom pricing. This plan targets production LLM deployments and combines agent security red-teaming, quality evaluation, integrations and collaboration, and hybrid deployment options. It also lists dedicated support with SLAs, with technical onboarding and consulting available optionally. On-premise, private cloud, and SaaS are deployment choices; Giskard says its team can help install Hub on-premise for sensitive workloads. The plan is the better fit when production controls, tailored evaluation, or remediation support matter, but custom pricing means teams should weigh those needs against their budget before committing.

Platforms

Giskard supports API, Linux, self-hosted, and web environments. The free plan’s local deployment and the Enterprise options spanning on-premise, private cloud, and SaaS give teams flexibility over where to run assessments. The Hub’s API requirement remains the key constraint on which agents it can assess.

Who it's for

Giskard is a strong fit for teams responsible for conversational AI who need to check security and quality before release, particularly those that value ranked findings, structured scenarios, and a described path from issue qualification to retesting. It also suits developers who want local behavioral tests in pytest. It is less suitable when the target is not an API-accessible conversational text agent, or when a team only needs a lightweight evaluation tool and does not need Giskard’s remediation and Enterprise controls.

Pros and cons

  • Pros: The assessment report prioritizes vulnerabilities and gives a deploy-or-fix verdict, helping teams decide what to address first.
  • Pros: The described ticketing and retesting process links findings to remediation rather than stopping at detection.
  • Pros: Local open-source tests and provider SDK integrations can fit into developer workflows, including pytest.
  • Cons: Hub assessments require a conversational text-to-text agent accessible through an API, narrowing the eligible systems.
  • Cons: The free plan provides basic scans and correctness-based RAG reporting, while advanced probes, evaluation tools, and production controls are Enterprise features.
  • Cons: Enterprise uses custom pricing, so production teams must assess cost against the value of its support and controls.

Alternatives

Consider Promptfoo if you want a free community tier with 10k red-team probes per month, all LLM evaluation features, broad provider integrations, and local or self-hosted operation. DeepEval is a fit for teams seeking a free, Apache 2.0-licensed open-source evaluation framework and local or CI/CD test runner. For a freemium platform with a free tier capped at 3 seats, one workspace, 10k logs per month, and three-day retention, consider Maxim AI.

Braintrust offers a free Starter plan with 1 GB of processed data, 10,000 scores, 14-day retention, and unlimited users, projects, and datasets. Parea AI offers a free plan capped at two team members, 3k logs per month with one-month retention, and 10 deployed prompts. Galileo is another freemium option, with a Pro plan at $100 a month billed yearly and 50,000 traces per month. For free public model comparisons, Arena (formerly Chatbot Arena) offers head-to-head battles and a leaderboard, subject to a daily limit that resets after 24 hours. Whisper is a free, MIT-licensed code-and-weights package that can be installed locally with Python and ffmpeg.

For broader category comparisons, browse AI Red Teaming Tools, AI Agent Evaluation Tools, AI Security Testing Tools, LLM Evaluation Tools, AI LLM Evaluation Tools, and LLM Observability Tools.

Verdict

Choose Giskard if you need structured security and quality assessments for API-accessible conversational agents, and value a process that carries findings through prioritization and retesting. Its free local library is a practical starting point; production teams gain the strongest case from the Enterprise probes, evaluation options, deployment flexibility, and support. Look elsewhere if your agent falls outside the supported format or if custom-priced Enterprise capabilities are more than your evaluation needs require.

The verdict

Giskard combines agent vulnerability assessments with structured findings and a described remediation process. Check that the supported agent type and plan fit your deployment and evaluation needs.

Questions people ask

Is Giskard free?

Yes. The Free plan costs nothing and includes the open-source library, local deployment, basic scans, best-effort maintenance and community support.

Where can Enterprise be deployed?

Enterprise lists on-premise, private cloud and SaaS deployment. Giskard also says its team can help install Hub on-premise for sensitive workloads.

What agents does Hub support?

Hub supports conversational text-to-text agents that are accessible through an API endpoint and is described as a black-box testing tool.

Which integrations are documented?

Giskard Checks documents native SDK integrations for OpenAI, Google/Gemini, Anthropic and Azure, with optional LiteLLM support for other providers.

Does it support human review and custom metrics?

Yes. Enterprise lists customizable metrics and human-in-the-loop review interfaces, along with domain-specific datasets and RAG quality metrics.

What security and compliance features are listed?

Giskard lists encryption at rest and in transit, data residency and isolation in the EU or US, access controls, audit trails, identity-provider integration, a zero-training policy, GDPR, SOC 2 Type II and HIPAA compliance.

What the maker publishes

API and SDK
Giskard publishes an official Python SDK for the Giskard Hub API in its public GitHub organization. source
Assessment report
After an assessment, Giskard provides a structured report with a deploy-or-fix verdict, ranked vulnerabilities, functional scenarios, and a Giskard Label if the agent passes. source
Behavioral checks
The Python library provides pytest-native scenarios and checks that can be written in Python or natural-language rules judged by an LLM. source
Black-box operation
The Hub tests an agent without requiring knowledge of its internal components or access to end-user production data. source
Collaboration
Hub includes shared workspaces, annotation workflows, role-based access control, dataset versioning and configurable alerting. source
Company
Giskard says it was founded in Europe in 2021 by co-founders with experience building AI systems at Dataiku and Thales. source
Company founding
LinkedIn lists the company as founded in 2021 and headquartered in Paris. source
Company size
LinkedIn lists Giskard as privately held with 11–50 employees. source
Compliance
The site states end-to-end encryption at rest and in transit and lists GDPR, SOC 2 Type II, and HIPAA compliance. source
Custom evaluations
Professional includes domain-specific dataset generation, fine-grained RAG metrics, customizable checks and interoperability with other evaluation platforms. source
Deployment
The Enterprise plan lists on-premise, private cloud, and SaaS deployment options, and the FAQ says its team can help install Hub on-premise for sensitive workloads. source
Documentation depth
Documentation includes installation, checks, scans, API references, migration guidance, examples, a glossary and community Discord support. source
Enterprise probes
The Professional offering lists more than 50 automated adversarial probes, including multi-turn attacks. source
Enterprise security controls
The Enterprise plan lists SSO, role-based access controls, versioning with audit trails, and scheduled email alerts. source
Evaluation features
The Enterprise plan lists domain-specific evaluation dataset generation, fine-grained RAG quality metrics, customizable metrics, and human-in-the-loop review interfaces. source
Founded
2021 source
Funding
Giskard says it raised a €1.5 million 2022 round led by Elaia with Bessemer and angel participation, followed by a €3 million European Commission investment in 2023. source
Headquarters
Europe source
Hub coverage
Giskard Hub supports text-processing agents including RAG chatbots and JSON-based agent workflows through an API endpoint. source
Human review
Professional provides collaborative human-in-the-loop review interfaces for evaluation results. source
Integrations
Giskard Checks documents native SDK integrations for OpenAI, Google/Gemini, Anthropic, and Azure, with optional LiteLLM support for other providers. source
Integrations
The earlier product description lists compatibility with Hugging Face, MLflow, Weights & Biases, PyTorch, TensorFlow and LangChain. source
License
The public giskard-oss repository is marked Apache-2.0 licensed. source
Limit: beta status
The April 2, 2026 v3 announcement labels the rewritten OSS library pre-release beta and says its APIs are still being refined. source

Checked against the maker's pages on 21 Oct 2026.