FD skill-vetter
Security-first skill vetting for AI agents. Use BEFORE installing any skill from ClawHub, GitHub, or other sources. Checks for red flags, dangerous patterns, permission scope, and suspicious code. Protects the agent and user from malicious skills.
As a process D 47/100 · Unfinished process — weak spots: when it triggers, inputs and preconditions, failures and branches
What is at stake
The skill contains fragments that, in the wrong hands, cost money or data. Below: what the installer risks and what the author should do.
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.
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
- Address the high-severity findings: each costs 18 safety points. If one is a false positive, add the rule id to guard.allow in spec.yaml.
- 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 · 24
-
high Dangerous commands
cmd-destructive-fsSKILL.md:56Destructive filesystem command (wipes root/home/drive) (documentation of a security skill)rm -rf / # Delete everything
security skill -
high Dangerous commands
cmd-destructive-fsSKILL.md:57Destructive filesystem command (wipes root/home/drive) (documentation of a security skill)rm -rf ~ # Delete home directory
security skill -
high Dangerous commands
cmd-destructive-fsSKILL.md:58Destructive filesystem command (wipes root/home/drive) (documentation of a security skill)rm -rf /* # Delete all files
security skill -
high Dangerous commands
cmd-destructive-fsSKILL.md:59Destructive filesystem command (wipes root/home/drive) (documentation of a security skill):(){ :|:& };: # Fork bombsecurity skill
Medium and low: 20
-
medium Exfiltration
exfil-read-secret-filesreferences/patterns.md:151Reads credential / secret files (documentation of a security skill)open('~/.ssh/id_rsa', 'r')security skill -
medium Dangerous commands
cmd-pipe-to-shellSKILL.md:32Downloads and executes remote code from an unrecognised host (pipe to shell) (documentation of a security skill)curl ... | bash # Remote code execution
security skill -
medium Dangerous commands
cmd-pipe-to-shellSKILL.md:33Downloads and executes remote code from an unrecognised host (pipe to shell) (documentation of a security skill)curl ... | sh # Remote code execution
security skill -
medium Dangerous commands
cmd-pipe-to-shellSKILL.md:34Downloads and executes remote code from an unrecognised host (pipe to shell) (documentation of a security skill)wget ... -O - | sh # Remote code execution
security skill -
medium Exfiltration
exfil-read-secret-filesSKILL.md:69Reads credential / secret files (documentation of a security skill)cat ~/.ssh/id_rsa # SSH key access
security skill -
low Risky intent
intent-offensive-securityreferences/patterns.md:35Offensive-security / dual-use content (legitimate for authorised testing; review intended use)### 2. Privilege Escalation
-
low Dangerous commands
cmd-privilegereferences/patterns.md:43Privilege escalation / world-writable permissions (documentation of a security skill)sudo rm -rf /var/log/*
security skill -
low Dangerous commands
cmd-privilegereferences/patterns.md:46Privilege escalation / world-writable permissions (documentation of a security skill)chmod 777 /etc/passwd
security skill -
low Dangerous commands
cmd-eval-dynamicreferences/patterns.md:90Dynamic code execution from decoded/untrusted input (documentation of a security skill)exec(base…ode("aW1w…nKQ=="))security skill -
low Dangerous commands
cmd-eval-dynamicreferences/patterns.md:91Dynamic code execution from decoded/untrusted input (documentation of a security skill)eval(compile(open('hidden.py').read(), 'hidden.py', 'exec'))security skill -
low Dangerous commands
cmd-pipe-to-shellscripts/vet_skill.py:77Downloads and executes remote code from an unrecognised host (pipe to shell) (detector / deny-list definition; string literal in code, not executed)(r'curl\s+.*\|\s*sh', "Remote code execution via curl | sh"),
detectorcode literal -
low Dangerous commands
cmd-destructive-fs-scopedscripts/vet_skill.py:97Destructive command on a partition / current directory (normal in admin docs; verify the target) (detector / deny-list definition; string literal in code, not executed)(r'rm\s+-rf\s+\*', "Dangerous rm -rf * pattern"),
detectorcode literal -
low Risky intent
intent-offensive-securitySKILL.md:39Offensive-security / dual-use content (legitimate for authorised testing; review intended use)**Privilege Escalation:**
-
low Dangerous commands
cmd-privilegeSKILL.md:42Privilege escalation / world-writable permissions (documentation of a security skill)chmod 777 ... # Overly permissive
security skill -
low Risky intent
intent-offensive-securitySKILL.md:51Offensive-security / dual-use content (legitimate for authorised testing; review intended use)nc -e /bin/sh ... # Reverse shell
-
low Dangerous commands
cmd-shell-rcSKILL.md:70Writes to a shell startup file (documentation of a security skill)cat ~/.bashrc # Shell config access
security skill -
low Dangerous commands
cmd-eval-dynamicSKILL.md:96Dynamic code execution from decoded/untrusted input (documentation of a security skill)exec(base…ode(...)) # Executing hidden code
security skill
A further 3 matches are quotations in this security skill's documentation and are not counted as findings.
Files scanned: 4. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
✓ No remarks against the Agent Skills spec
Process rating: all ten parameters 47/100
- 0Inputs and preconditions. Does not say what the process needs to start
- 0Failures and branches. Linear process with no failure handling
- 0Progress reporting. Says nothing while it works
- 20When it triggers. No condition that starts the skill
- 30Running it twice. 3 mutating operations with no state check
- 40Consistency. Frontmatter name (skill-vetter) differs from the folder (vivi-skill-vetter)
- 60Tools and files. Uses tools (web, git, python) that frontmatter does not declare
- 60Result and completion. Output format stated, no completion criterion
- 100Steps. 51 steps
- 100Execution cost. Instruction body is 2372 tokens
- medium Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
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
- +4Description does not say when NOT to use the skill (false activations)
- -213 emoji in the instructions: noise for the model
- -41 reference files, but SKILL.md never points to them: the model will not open them
- +1No license
- +2Single-language instructions
- +3Description length 247: enough signal without eating the budget
- +4Structure: 26 headings
- +3Step-by-step instructions: 51 items
- +3Output format is stated explicitly
- +4Has examples (18 code blocks)
- +3All 1 scripts are documented
Quality base 70; lint remarks subtract, signals add up to 100. Result: 84.