competition-agent-cloud

1# Competition Agent Cloud
2 
3Use this skill only as a downstream specialization after `$ctf-sandbox-orchestrator` is already active and has established sandbox assumptions, node ownership, and evidence priorities. If that has not happened yet, return to `$ctf-sandbox-orchestrator` first.
4 
5Use this skill when the challenge path is driven by prompt-to-tool execution, retrieval and memory boundaries, deployment drift, or build and release provenance.
6 
7Reply in Simplified Chinese unless the user explicitly requests English.
8 
9## Quick Start
10 
111. Decide whether the dominant path is agentic or infrastructure-driven.
122. Map one minimal control chain: untrusted input -> visible context -> tool or deployment side effect.
133. Distinguish checked-in intent from live runtime truth.
144. Keep prompts, tool args, manifests, mounts, and provenance steps in compact evidence blocks.
155. Reproduce the exploit or misconfiguration with minimal context and minimal instrumentation.
16 
17## Workflow
18 
19### 1. Agent And Prompt Injection
20 
21- Treat prompts, tool schemas, retrieved chunks, planner notes, memory files, and handoffs as challenge artifacts.
22- Prove one minimal chain from untrusted content to model-visible instruction to tool side effect.
23- Distinguish claimed capability from runtime-exposed capability.
24 
25### 2. Cloud, Containers, And CI/CD
26 
27- Split build-time, deploy-time, and runtime.
28- Reconcile compose or kube manifests with live mounts, env, logs, and traffic.
29…