Browse categories
Explore
Fiverr Pro
English
$
USD
I will audit your ai agent for prompt injection and ai act readiness
Mustaqeez Ahmad
About this gig
AI agent security audit for LLM apps, chatbots and voice agents. I test prompt injection, jailbreaks, data leakage, tool abuse, and AI Act readiness.
You get a practical technical report with severity, proof-of-concept prompts where safe, evidence, and remediation steps. This is a technical readiness review, not legal advice or formal compliance certification.
What you get
- Prompt injection and jailbreak testing(OWASP LLM Top 10)
- Data exfiltration and PII leakage assessment
- Tool/function abuse and privilege escalation tests
- System prompt extraction attempts
- AI Act readiness review(risk classification support, transparency checks, Annex IV-style evidence starter where relevant)
- GDPR / data protection technical gap notes (where applicable)
- Written report with severity ratings + remediation playbook
Frameworks I work with
LangChain, LangGraph, CrewAI, AutoGen, VAPI/Bland/Retell, OpenAI Assistants API, Anthropic agents, Claude Code projects, n8n AI agents, custom FastAPI agents.
Message me on Fiverr before ordering with your AI system, frameworks, data types, tools/actions, and access limits. I will recommend the right package or send a custom Fiverr offer. No legal advice included
Get to know Mustaqeez Ahmad
Mustaqeez Ahmad
I'm an expert in generative AI mobile app and web SaaS development
- FromPakistan
- Member sinceJun 2016
- Last delivery1 year
Languages
Urdu, English
I have a Masters's degree in Computer Science and 10 years plus experience in development. I'm CEO at Software & Technology Company "CodeDeskStudio" and managing a small team. We specialize in AI SaaS development.
As a team, we have good UI UX design skills to make top-level apps UI.
My Portfolio
FAQ
What's prompt injection and why does it matter?
Prompt injection is when an attacker manipulates an LLM's instructions through user input - leaking system prompts, exfiltrating data, or hijacking tools. It's #1 on the OWASP LLM Top 10. If your AI handles any sensitive data, you're exposed.
Do I need this if I'm just a small SaaS or startup?
If your product serves EU users, handles sensitive workflows, or may fall into a high-risk or transparency category, you should start readiness work early. I provide a technical gap review and remediation roadmap. This is not legal advice; for formal compliance decisions, use qualified counsel
What frameworks can you audit?
LangChain, LangGraph, CrewAI, AutoGen, OpenAI Assistants/Agents API, Anthropic Claude agents, VAPI/Bland/Retell voice agents, n8n AI workflows, custom FastAPI/Python agents. Tell me your stack.
What does the report include?
Executive summary, prioritized vulnerability list with severity, proof-of-concept exploit prompts where safe to include, remediation playbook, and Premium AI Act readiness notes with an Annex IV-style evidence starter. All written, traceable, and practical.
Will you sign an NDA?
Yes. I sign before any system access. For enterprise engagements I can also work under your standard MSA. Send your NDA before kickoff
How long does an audit take?
Basic: 5 days. Standard: 7 days. Premium: 14 days (more for very large agent ecosystems). Enterprise scope is custom-quoted.
Do you also fix the issues you find?
I deliver remediation guidance in the report. If you want me to implement the fixes, that's a separate engagement quoted after the audit.
Can you help with EU AI Act classification?
Yes - Premium includes a practical risk-classification worksheet and Annex IV-style starter documentation where relevant. I do not provide legal opinions or conformity assessment; I help your technical team prepare evidence and fix gaps.
What if you find nothing wrong?
That's a passing audit and you get a clean report you can show clients, investors, or auditors. Worth the price for the documentation alone.
Do you offer ongoing security retainers?
Yes - monthly retainers from $700/mo for continuous security monitoring, prompt updates, and vulnerability tracking. Discuss after the first audit.
