sec-skill-network: Skill script makes network calls

What it catches#

Skill script makes network calls.

A script in a skill folder that makes network calls (curl, wget, requests, fetch and the like). The note lists the hosts written in the script, or says the host is not written literally, so a reviewer can judge them. Markdown in the skill folder is not checked for this. Needs --include skills. Not a problem by itself: many skills call an API.

Why it matters#

Lists the hosts so a reviewer can judge them. Not a problem by itself.

Bad example#

# scripts/fetch-docs.sh
curl -s https://api.example.invalid/v1/docs
curl -s https://203.0.113.9/x

The second host is a bare IP with no mention anywhere in the skill. That is the one to ask about.

Good example#

# scripts/fetch-docs.sh
# Reads the public docs index from api.example.invalid (see SKILL.md, "Network").
curl -s https://api.example.invalid/v1/docs

This still gets the note, since every network call does. The difference is that the one host is the API the skill is for, and SKILL.md says so.

How to fix#

Nothing, if the hosts are the ones the skill is for. Name them in SKILL.md so the next reviewer does not have to grep. Remove calls to hosts you cannot explain.

Suppress or configure#

sec-skill-network is a security rule, so it cannot be suppressed with a comment in the file: a suppression comment is itself a hidden comment, and an attacker who can add a payload can add one next to it. A comment that tries gets a suppression-not-allowed warning.

Turn it off, or change its severity, in .asitisrc.json, which shows up in review:

{ "rules": { "sec-skill-network": "off" } }

Or accept the findings you have today in a baseline, so only new ones fail the build:

asitis lint --format json --write-baseline > .asitis-baseline.json
asitis lint --baseline .asitis-baseline.json

Source#

Snyk: ToxicSkills, malicious AI agent skills, read on 14 October 2026.

sec-skill-network