Skillv1.0.0

ClawScan security

Nsg Firewall Auditor · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

BenignMar 4, 2026, 11:04 AM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill is an instruction-only Azure NSG/Firewall auditing guide that asks users to paste exported az CLI JSON outputs for analysis; its requirements and instructions are consistent with that purpose.
Guidance: This skill is instruction-only and appears coherent, but take these precautions before using it: (1) Run the az commands locally yourself and review the JSON output — do not paste any credentials, tokens, or connection strings. (2) If you must share data, redact or replace sensitive identifiers and public IPs when possible, or share a sanitized example. (3) Use least-privilege roles locally (Reader is sufficient for many exports; Network Contributor is only needed for some effective-rule queries). (4) The skill will produce recommended CLI commands and tightened JSON — review those before applying them in your environment. (5) Note the skill lists a price and 'pack' metadata in its header; this is informational and not required to run the guidance. If you need the auditor to run commands directly against your subscription, prefer a vetted tool or grant temporary, audited access rather than pasting secrets into chat.

Purpose & Capability: okThe name/description match the instructions: the skill asks users to provide az CLI exports (NSG lists, effective rules, firewall policies) and describes checks and remediation. It does not request unrelated credentials, binaries, or access.
Instruction Scope: noteSKILL.md stays within scope (it instructs the agent to analyze user-provided exports and not to run CLI against the user's account). Minor inconsistency: the front-matter lists 'tools: claude, bash' though the body emphasizes that the skill will not execute Azure CLI — this is likely informational but could confuse less technical users. The skill explicitly warns to confirm exported data contains no credentials before processing.
Install Mechanism: okNo install spec and no code files — instruction-only, so nothing is written to disk or downloaded. This is the lowest-risk install model and matches the stated behavior.
Credentials: okNo environment variables, no credentials, and no config paths are requested. The skill requests exported CLI output and recommends minimum RBAC roles for running those CLI commands locally; those requirements are proportionate to the auditing task.
Persistence & Privilege: okalways is false and model invocation is allowed (platform default). The skill does not request persistent presence or modification of other skills or system-wide settings.