Week-by-week plan
| Week | Phase | Key activities | Output |
|---|---|---|---|
| Weeks 1–2 | Threat Modelling | Map model inventory, data flows, agent tools and trust boundaries against OWASP LLM Top 10 and MITRE ATLAS tactics. | Signed-off threat modelling pack and gate review |
| Weeks 3–4 | Adversarial Testing | Prompt injection (direct and indirect), jailbreaks, system-prompt leakage, training-data extraction, model DoS, RAG corpus poisoning and tool/function-call abuse. | Signed-off adversarial testing pack and gate review |
| Weeks 5–6 | Platform & Pipeline | Pentest the surrounding stack — vector DBs, inference endpoints, fine-tuning pipelines, MLOps tooling, model registries and identity boundaries. | Signed-off platform & pipeline pack and gate review |
| Weeks 7–8 | Reporting | Risk-ranked report with PoCs, screen captures, repeatable payload library and CVSS-aligned scoring. | Signed-off reporting pack and gate review |
| Weeks 9–12 | Re-test | Validation of remediated findings and tuning of guardrails, evaluators and detection rules. | Signed-off re-test pack and gate review |
Gantt-style timeline titled "12-week delivery plan" over 12 Weeks with 5 phases: Threat Modelling from Week 1 to 2; Adversarial Testing from Week 3 to 4; Platform & Pipeline from Week 5 to 6; Reporting from Week 7 to 8; Re-test from Week 9 to 12.