Skip to content
Get packs

Competition Agent Cloud

Guides an agent through CTF-style analysis of AI-agent tool flows, cloud/container and CI/CD challenges.

Source Download skill
Repository
zhaoxuya520/reverse-skill
Stars
43,969
Licence
MIT
Files
3 files

What it does

This is a specialist skill for a CTF (capture-the-flag) security-competition workflow. Once an agent is already running the parent orchestrator skill, it helps trace how untrusted input such as prompts or retrieved documents can drive an AI agent's tools, and how deployment manifests, mounted secrets, and build/release artifacts differ from what is actually running. It tells the agent to keep compact evidence (prompts, tool arguments, manifests, hashes) and to reply in Simplified Chinese unless English is requested. It also points to other more specific competition skills for topics like prompt injection, supply chain, SSRF, or container escape.

Great for

  • Students or players working on AI-agent and cloud security CTF challenges inside a sandbox
  • Security researchers reviewing prompt-to-tool flows and deployment drift in a lab setting
  • Teams already using the ctf-sandbox-orchestrator skill who want this downstream specialization

Skip it if

  • Attacking production systems or third-party infrastructure: the skill is written for sandboxed competition challenges and gives no authorization or scope checks
  • Use without the parent ctf-sandbox-orchestrator skill: its own instructions say to return to the orchestrator first, so it will not work well standalone
  • Non-security tasks: it contains no general guidance on building agents, cloud deployment, or CI/CD

Try asking

  • “Trace how a poisoned retrieved document in this CTF challenge leads an agent to call a tool with attacker-chosen arguments”
  • “Compare the docker-compose manifest with the live container mounts and environment for this sandbox challenge”
  • “Check whether the artifact hash in the CI output matches what the runtime consumer actually loaded”
Step-by-step guide: how to use Competition Agent CloudInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 3 original files, the MIT licence and an attribution note.

    competition-agent-cloud.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip competition-agent-cloud.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip competition-agent-cloud.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Install the parent ctf-sandbox-orchestrator skill first, since this one is designed to be routed to only after it has set sandbox assumptions
  • State the sandbox and challenge scope explicitly so the agent keeps analysis within it
  • Provide concrete artifacts (prompts, manifests, logs) rather than vague descriptions to get useful evidence blocks

What's inside

  • agents/openai.yaml
  • references/agent-cloud.md
  • SKILL.md

Redistributed unmodified from zhaoxuya520/reverse-skill under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.