What problem does it solve? CTF challenges involving AI agents, cloud, containers, and CI/CD often hide the flag in the gap between checked-in intent and live runtime truth. This Skill provides a structured methodology for proving minimal exploit chains across prompt injection, deployment drift, and supply-chain provenance under sandbox assumptions. ## Core Features & Use Cases - Agentic Path Analysis: Maps instruction layers, retrieval layers, memory, tool gates, and side effects to prove one minimal chain from untrusted content to tool side effect. - Cloud and Container Reconciliation: Compares compose or Kubernetes manifests against live mounts, environment variables, logs, and traffic across build-time, deploy-time, and runtime. - Supply-Chain Provenance Tracing: Tracks the chain from source to dependency resolution, build, publish, and runtime consumer, focusing on version drift and registry pulls. - Use Case: During a CTF, you face an AI agent whose retrieved documents influence tool calls backed by a containerized service. Use this Skill to prove the prompt-to-tool chain, reconcile the manifest with live mounts, and package compact evidence blocks. ## Quick Start Ask the agent to analyze the challenge's prompt-to-tool flow and reconcile the deployment manifests with live runtime state after the sandbox orchestrator has established assumptions.
