CB analyzing-malware-behavior-with-cuckoo-sandbox
Detonate malware samples in Cuckoo Sandbox to observe runtime behavior — process creation, file system and registry changes, network communications, and API calls — and generate behavioral reports for classification and IOC extraction. Use when a sample has passed static triage and needs dynamic/behavioral analysis, when mapping a full infection chain, or when building YARA/behavioral signatures from observed sandbox activity.
Detonate malware samples in Cuckoo Sandbox to observe runtime behavior — process creation, file system and registry changes, network communications, and API…
As a process B 65/100 · Nearly there — weak spots: 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 skill contains commands that delete files, rewrite disks or execute code fetched from the network. The agent may run them without asking if it believes the instructions require it.
Replace destructive commands with safe equivalents that ask for confirmation, scope them to one folder, and stop piping curl into a shell: pin a version and a checksum.
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 · 5
✓ No critical or high findings
Medium and low: 5
-
medium Dangerous commands
cmd-persistenceSKILL.md:287Persistence mechanism (cron / launchd / scheduled task / autorun registry) (documentation of a security skill)[SET] HKCU\Software\Microsoft\Windows\CurrentVersion\Run\WindowsUpdate = "C:\Windows\System32\svchost_fake.exe"
security skill -
low Risky intent
intent-offensive-securitySKILL.md:44Offensive-security / dual-use content (legitimate for authorised testing; review intended use)**Do not use** when the sample is a known ransomware variant that may spread via network shares in a misconfigured sandbox; verify network isolation first.
-
low Risky intent
intent-offensive-securitySKILL.md:237Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (detector / deny-list definition)- **Cuckoo Sandbox**: Open-source automated malware analysis system providing behavioral reports, network captures, and memory dumps
detector -
low Risky intent
intent-offensive-securitySKILL.md:238Offensive-security / dual-use content (legitimate for authorised testing; review intended use)- **InetSim**: Internet services simulation suite providing fake HTTP, DNS, SMTP, and other services for isolated malware analysis networks
-
low Risky intent
intent-offensive-securitySKILL.md:247Offensive-security / dual-use content (legitimate for authorised testing; review intended use)**Context**: Static analysis reveals a packed executable with minimal imports and high entropy. The sample needs sandbox execution to observe unpacking, payload delivery, and C2 establishment.
Files scanned: 4. 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 65/100
- 0Failures and branches. Linear process with no failure handling
- 30Running it twice. 9 mutating operations with no state check
- 50When it triggers. No condition that starts the skill
- 60Tools and files. Uses tools (web) that frontmatter does not declare
- 60Result and completion. Output format stated, no completion criterion
- 70Inputs and preconditions. Inputs and preconditions are listed
- 100Steps. 32 steps
- 100Consistency. Name and required fields are in place
- 100Execution cost. Instruction body is 2715 tokens
- 100Progress reporting. Reports progress
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
- -4Absolute local paths (C:\Users, /home/…): not portable
- -2localhost URLs: will not work for another user
- -41 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 430: enough signal without eating the budget
- +4Structure: 16 headings
- +3Step-by-step instructions: 32 items
- +3Output format is stated explicitly
- +4Has examples (8 code blocks)
- +1License stated
Quality base 70; lint remarks subtract, signals add up to 100. Result: 76.