CC escaping-containers-to-host
Exploits privileged pods, host mounts, runC CVEs, and exposed Docker sockets to break out of a container and reach the underlying host during an authorized container-security assessment. Use when executing an approved breakout test, demonstrating the real impact of a privileged or hostPath workload, or validating that escape mitigations actually hold. Keywords: container breakout, privileged, hostPath, docker.sock, runC CVE-2019-5736, CVE-2024-21626, release_agent, nsenter. Do not use for defensive detection of these techniques - use detecting-container-escape-attempts.
Exploits privileged pods, host mounts, runC CVEs, and exposed Docker sockets to break out of a container and reach the underlying host during an authorized…
As a process C 55/100 · Has gaps — weak spots: result and completion, failures and branches, running it twice
What is at stake
Medium-severity findings: the skill is probably honest, but read what alarmed the scanner.
Below is the worst case for this category. The finding here is medium: the guard saw a sign, not a proof.
The instructions or scripts send data out: environment variables, keys, file contents, chat history. You may never notice, because the agent performs the upload as "part of the task".
If the upload is not needed for the task, remove it: catalogs flag such skills and delist them. If it is needed, name the destination explicitly, say what leaves the machine, and give the user a switch.
How to improve
- Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 8
✓ No critical or high findings
Medium and low: 8
-
medium Exfiltration
exfil-read-secret-filesSKILL.md:137Reads credential / secret files (documentation of a security skill)cat /etc/shadow >> /output
security skill -
low Risky intent
intent-offensive-securityreferences/standards.md:7Offensive-security / dual-use content (legitimate for authorised testing; review intended use)| T1611 | Escape to Host | Privilege Escalation | The core technique: breaking container isolation to execute on the host kernel/namespace. |
-
low Risky intent
intent-offensive-securityreferences/standards.md:10Offensive-security / dual-use content (legitimate for authorised testing; review intended use)| T1068 | Exploitation for Privilege Escalation | Privilege Escalation | runC CVE exploitation (CVE-…626, CVE-…133/52565/52881). |
-
low Risky intent
intent-offensive-securitySKILL.md:49Offensive-security / dual-use content (legitimate for authorised testing; review intended use)- During an authorized container or Kubernetes penetration test after obtaining initial code execution inside a container or pod
-
low Risky intent
intent-offensive-securitySKILL.md:65Offensive-security / dual-use content (legitimate for authorised testing; review intended use)# CDK - zero-dependency K8s/container pentest toolkit
-
low Risky intent
intent-offensive-securitySKILL.md:83Offensive-security / dual-use content (legitimate for authorised testing; review intended use)| T1611 | Escape to Host | Privilege Escalation |
-
low Risky intent
intent-offensive-securitySKILL.md:86Offensive-security / dual-use content (legitimate for authorised testing; review intended use)| T1068 | Exploitation for Privilege Escalation | Privilege Escalation |
A further 1 matches are quotations in this security skill's documentation and are not counted as findings.
Files scanned: 5. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
- note
frontmatter-keyunknown frontmatter key "domain" - note
frontmatter-keyunknown frontmatter key "subdomain" - note
frontmatter-keyunknown frontmatter key "nist_csf" - note
frontmatter-keyunknown frontmatter key "mitre_attack"
Process rating: all ten parameters 55/100
- 0Result and completion. Does not say what the result is
- 0Failures and branches. Linear process with no failure handling
- 0Progress reporting. Says nothing while it works
- 30Running it twice. 4 mutating operations with no state check
- 50When it triggers. No condition that starts the skill
- 60Tools and files. Uses tools (bash, web, node) that frontmatter does not declare
- 70Inputs and preconditions. Inputs and preconditions are listed
- 100Steps. 24 steps
- 100Consistency. Name and required fields are in place
- 100Execution cost. Instruction body is 3163 tokens
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +5Description has no quoted example phrases that should trigger the skill
- +3Output format is not stated: the model decides each time
- -2localhost URLs: will not work for another user
- -42 reference files, but SKILL.md never points to them: the model will not open them
- -31 of 1 scripts are never mentioned in SKILL.md
- +2Single-language instructions
- +4Description says when NOT to use the skill
- +3Description length 576: enough signal without eating the budget
- +4Structure: 17 headings
- +3Step-by-step instructions: 24 items
- +4Has examples (8 code blocks)
- +1License stated
Quality base 70; lint remarks subtract, signals add up to 100. Result: 77.