AC skill-guard
Scan ClawHub skills for prompt injection and malicious content using Lakera Guard before installing them. Run automatically when the user asks to install a skill, or on-demand to audit any skill by slug or search query.
As a process C 64/100 · Has gaps — weak spots: inputs and preconditions, consistency
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 · 2
✓ No critical or high findings
Medium and low: 2
-
medium Exfiltration
exfil-webhook-urlINTEGRATION.md:83Webhook / callback URL commonly used for exfiltration (verify the destination) (documentation of a security skill)# Use the https://*.trycloudflare.com URL it gives you
security skill -
low Exfiltration
exfil-secret-in-urlscripts/scan.sh:89Credential passed in a URL query string (normal for some APIs — verify the host is the intended service) (placeholder value)"https://api.apify.com/v2/acts/${ACTOR_ID}/runs?token=…&webhooks=…" \placeholder
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 64/100
- 0Inputs and preconditions. Does not say what the process needs to start
- 40Consistency. Frontmatter name (skill-guard) differs from the folder (skill-guard-actor)
- 60Tools and files. Uses tools (bash, web) that frontmatter does not declare
- 60Result and completion. Output format stated, no completion criterion
- 65Failures and branches. 3 branches
- 70When it triggers. States when to use, but not when not to
- 100Steps. 11 steps
- 100Execution cost. Instruction body is 1610 tokens
- 100Running it twice. Mutating operations check current state
- 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
- +4Description does not say when NOT to use the skill (false activations)
- +1No license
- +2Single-language instructions
- +3Description length 219: enough signal without eating the budget
- +4Structure: 19 headings
- +3Step-by-step instructions: 11 items
- +3Output format is stated explicitly
- +4Has examples (9 code blocks)
- +3All 1 scripts are documented
Quality base 70; lint remarks subtract, signals add up to 100. Result: 90.