- Home
- AI Security Testing Tools
- AIRTA Red Team
AIRTA Red Team
What it is
AIRTA Red Team is a free, self-hosted tool for authorized adversarial assessments of LLMs and applications built around them. It builds test suites from security playbooks, identifies target components and authentication, runs attacks, evaluates results, and exports reports. Listed playbooks cover OWASP LLM, OWASP Agent, MITRE ATLAS, jailbreaks, system prompt exfiltration, and custom cases. Strategies include zero-shot, multi-shot, few-shot, iterative, multimodal, jailbreak, and chain-of-thought testing. Targets can be reached through a Playwright browser runner or direct HTTP API transports, including file uploads and tool calls. The product can run in Docker locally, behind a firewall, in CI, or against external endpoints for authorized black-box assessments. Runs record prompts, uploads, responses, and tool calls in attack_log.json; pipeline_report.json summarizes categories and severity. Results can be exported to AIRTA Systems through its bulk import API for review and regression tracking. Users must supply their own LLM API key for test generation and assessment. Its scope is observable runtime behavior rather than source-code static analysis. The repository states that the tool is deprecated and no longer maintained.
For and against
For
- Supports OWASP LLM, OWASP Agent, and MITRE ATLAS playbooks
- Can run locally, behind a firewall, or in CI
- Captures prompts, uploads, responses, and tool calls
- Exports reports with category summaries and severity
Against
- Deprecated and no longer maintained
- Users must provide their own LLM API key
- Assessments do not perform source-code static analysis
The verdict
AIRTA Red Team offers self-hosted testing workflows and report exports for runtime assessments, with no license fee or subscription. Its deprecated, unmaintained status is an important limitation to consider.
Who it suits
It is aimed at red teams, whitehat testers, pentesters, and AppSec or MLsec teams assessing LLM behavior. Use it only to test systems for which you have explicit permission.
Plans and prices
Free
Full pipeline, web UI, playbooks and CLI · self-hosted use
Read 15 Oct 2026 · airtasystems.com
AIRTA Red Team
Open source; no license fee or subscription · Bring your own LLM API key
Read 4 Oct 2026 · airtasystems.com
On the record
- Assessment models
- Playbook-specific expert models and a judge model classify attacks as exploited or mitigated and assign severity by category.airtasystems.com · read 15 Oct 2026
- Assessment scope
- The product page says assessments focus on observable runtime behavior rather than source-code static analysis.airtasystems.com · read 4 Oct 2026
- Attack coverage
- Listed playbooks include OWASP LLM, OWASP Agent, MITRE ATLAS, jailbreak, system prompt exfiltration, and custom playbooks.airtasystems.com · read 4 Oct 2026
- Attack strategies
- The documented strategies include zero-shot, few-shot, multi-shot, jailbreak, multimodal, iterative and chain-of-thought testing.airtasystems.com · read 15 Oct 2026
- Authorization
- The product page says to test only systems for which the user has explicit permission.airtasystems.com · read 4 Oct 2026
- Authorized use
- The maker restricts use to systems for which the tester has explicit permission.airtasystems.com · read 15 Oct 2026
- Capterra pricing listing
- Capterra lists AIRTA Systems pricing as contact-vendor, with no free trial and no free version included in that listing.capterra.com · read 15 Oct 2026
- Company founders
- The about page says AIRTA Systems was founded in London by Cindy Ogidi and Bob Morel.airtasystems.com · read 15 Oct 2026
- Company founding year
- LinkedIn lists the company as founded in 2026.linkedin.com · read 15 Oct 2026
- Company legal name
- The product site footer identifies the maker as AIRTA Systems Ltd.airtasystems.com · read 15 Oct 2026
- Company size
- LinkedIn lists AIRTA Systems as having 11–50 employees.linkedin.com · read 15 Oct 2026
- Company type
- LinkedIn classifies AIRTA Systems as a public company in the technology, information and internet industry.linkedin.com · read 15 Oct 2026
- Deployment
- The maker documents local Docker deployment with python start.py, CLI operation with main.py, CI use, behind-firewall staging and authorized external black-box testing.airtasystems.com · read 15 Oct 2026
- Documentation
- The maker provides a support documentation page with article search and category filters, but the page currently shows a loading posts area rather than a visible Red Team manual.airtasystems.com · read 15 Oct 2026
- Editable suites
- Attack suites are JSON files containing playbooks, categories, prompts and optional multimodal payloads that can be edited in the web interface.airtasystems.com · read 15 Oct 2026
- Evidence
- Runs capture prompts, uploads, responses, and tool calls in attack_log.json and produce pipeline_report.json with category rollups and severity.airtasystems.com · read 4 Oct 2026
- Evidence files
- Runs capture prompts, uploads, responses and tool calls in attack_log.json and summarize results in pipeline_report.json.airtasystems.com · read 15 Oct 2026
- Execution
- Targets can be tested through a Playwright browser runner or direct HTTP API transports, including file uploads and tool calls.airtasystems.com · read 4 Oct 2026
