
Giskard
Source check found a deprecation noticeContinuous AI red teaming platform that detects vulnerabilities and safeguards AI agents.
Source-checked against the vendor's site Β· rebuilt from the vendor's own pagesProfile updated Link checked How we rate
In short
Giskard is a black-box AI red teaming platform that detects vulnerabilities like prompt injections. It offers a freemium model with open-source local deployment and professional automated scans.
Giskard operates as a black-box testing tool that does not require access to the internal components or production data of the AI agent.
Giskard detects risks such as harmful content, personal information disclosure, and prompt injections using proprietary research and risk-based test generation. Giskard provides a structured report with a deploy or fix verdict, ranking security vulnerabilities by criticality and assigning a Giskard Label if the agent passes. Giskard manages the remediation process by qualifying findings, opening tickets in the user's workflow, and re-running tests until fixes are confirmed.
The open-source version supports local Python deployment with bring-your-own-LLMs and covers 7 risk categories for agent security vulnerability detection. The Professional plan includes agent-specific automated vulnerability scans, audit reports, and over 50 automated adversarial probes including multi-turn attacks. The vendor says that Giskard offers secure deployment options including on-premise, private cloud, and SaaS, with data residency, a 0-training policy, and SOC 2, HIPAA, and GDPR compliance.
According to Giskard: Giskard operates as a black-box testing tool that does not require access to the internal components or production data of the AI agent. Its pages list this strength: Black-box testing capability that does not require access to model internals or user data. They also state this limitation: Reporting in the open-source version is limited to code.
β The AI Tool Bible editorial team
Pros
- β Black-box testing capability that does not require access to model internals or user data.
- β The vendor says that compliance with SOC 2, HIPAA, and GDPR standards.
- β Includes a remediation workflow that opens tickets and re-runs tests to confirm fixes.
- β Supports various AI agent types, including RAG chatbots and agentic workflows processing JSON.
Cons
- β οΈ Reporting in the open-source version is limited to code.
Facts
- Platforms
- web
- API available
- Yes (not source-checked)
- Free trial
- Yes (not source-checked)
- Open source
- Yes (license: Apache-2.0) (not source-checked)
- GitHub stars
- 5,949 (checked Nov 13, 2026) (not source-checked)
- Last GitHub push
- Nov 12, 2026 (not source-checked)
- First commit
- Mar 2022 (not source-checked)
- Company
- Giskard AI (not source-checked)
- Model used
- Multi-model (not source-checked)
- Package downloads (30 d)
- 12,622 Β· pypi giskard Β· checked Nov 12, 2026 (not source-checked)
- Hacker News
- 15 stories Β· top 114 points Β· last Jun 24, 2026 Β· checked Nov 12, 2026
- Earliest Wayback capture
- 2021-09-24 (earliest archived capture of the site, not a launch date)
- Headquarters
- France (Wikidata)
Where this tool sits
Use cases
Frequently asked
- How much does Giskard cost?
- Giskard operates on a freemium pricing model. The open-source version supports local Python deployment with bring-your-own-LLMs. Specific pricing details for the Professional plan are not documented in the provided information.
- Does Giskard require access to my AI model's internal code?
- No, Giskard operates as a black-box testing tool. It does not require access to the internal components or production data of the AI agent to perform its security assessments and vulnerability detection.
- What specific security risks does Giskard detect?
- Giskard detects risks including harmful content, personal information disclosure, and prompt injections. The open-source version covers 7 risk categories, while the Professional plan includes over 50 automated adversarial probes, including multi-turn attacks.
- What deployment options are available for Giskard?
- Giskard offers secure deployment options including on-premise, private cloud, and SaaS. It supports data residency, maintains a 0-training policy, and complies with SOC 2, HIPAA, and GDPR standards.
- How does Giskard handle remediation after finding vulnerabilities?
- Giskard manages remediation by qualifying findings, opening tickets in the user's workflow, and re-running tests until fixes are confirmed. It provides a structured report with a deploy or fix verdict and ranks vulnerabilities by criticality.