Skip to content
SkillModel-led676 tokens in use

competition-agent-cloud

A downstream CTF-sandbox skill for AI-agent, prompt-injection, MCP, cloud, container and CI/CD challenges, with a checklist for deployment truth and provenance.

44,996 stars on the repository MIT Licensezhaoxuya520/reverse-skill · CTF-Sandbox-Orchestrator/competition-agent-cloud

Skill label

measured from its files

What installing competition-agent-cloud costs

110tokens idle
676tokens in use
894tokens, up to

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Up to adds the files its instructions link. Counts are approximate, made offline with cl100k_base.

Model dependence Model-led

Script-ledMixedModel-led

Mostly instructions, so how well it works depends on the model that follows them.

This describes the skill's structure, how much of it is scripts. It does not say how well it works.

Touches

Nothing detected: no shell, network, file writes, git or credentials in its files.

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Good
  • Frontmatter12/15
  • Structure23/25
  • Examples3/15
  • Output10/15
  • Size15/15
  • Resources7/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Excellent

No findings.

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Good
  • Development activity96
  • Maintenance72
  • Community72
  • Maturity28
  • Momentum40

This skill is a specialization of the ctf-sandbox-orchestrator skill. It covers challenges where the path runs through prompt-to-tool execution, retrieval and memory boundaries, deployment drift, or build and release provenance.

The agent maps one minimal control chain from untrusted input to visible context to a tool or deployment side effect, and keeps checked-in intent apart from live runtime state. The skill states that it runs only after the orchestrator has set sandbox assumptions.

What You Get

  • A workflow split into agent and prompt-injection analysis, and cloud, container and CI/CD analysis
  • A reference checklist for the agentic path, the cloud and container path, and supply-chain provenance
  • A routing table that sends narrower questions to nine sibling competition skills
  • An evidence list covering prompt snippets, manifests, mounts, artifact hashes and CI steps

Common Use Cases

  • Analyzing how retrieved or untrusted content reaches a tool call in a CTF agent challenge
  • Comparing a checked-in compose or Kubernetes manifest with live mounts and environment
  • Tracing a build artifact from source through dependency resolution to its runtime consumer

What Makes It Unique

It acts as a switchboard for the agent and cloud corner of the competition set. It gives a short general method, then names the sibling skill to prefer for prompt injection, supply chain, queue workers, SSRF, request smuggling, cloud metadata, Kubernetes control plane, container runtime and container escape questions.

Under The Hood

The skill is instruction text only: a SKILL.md, one reference checklist and an agent configuration file. It ships no scripts and does not tell the agent to fetch pages or call APIs. The method itself does cover inspecting manifests, mounts, logs and traffic, so the agent’s own tools do that work inside the sandbox the orchestrator set up.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers