The Agent Skills Directory

[SAFE]: The skill implements a network diagnostic analyzer for local packet captures. It functions as a read-only tool that processes user-supplied files and generates a technical report on TCP performance, VPN negotiation, DNS, and TLS handshakes. All analysis logic is contained within the local Python script and the Scapy library. No data exfiltration, remote code execution, or credential harvesting patterns were detected. Findings across evaluated categories are as follows:
Read-only local analysis: The skill instructions and code (scripts/pcap_analyze.py) explicitly mandate read-only operations. It does not modify files or environment settings.
Local dependencies: The skill requires the scapy package, a well-known and standard library for packet manipulation. The scripts/requirements.txt specifies scapy>=2.5,<3.
Offline operation: The analysis script performs no network operations (no requests, urllib, or socket-based exfiltration). It processes data entirely in-memory from the provided local pcap file.
Safe command execution: Commands provided in SKILL.md are limited to running the local analyzer script with parameters for filtering or outputting results to a Markdown file.
Privilege and Persistence: No attempts to escalate privileges, use sudo, or establish persistence (cron, startup scripts) were found. The remediation advice for MTU/buffer issues suggests manual sudo commands for the user to run if they choose, which is standard documentation practice.
Metadata and Injections: Metadata is consistent with the skill's purpose. No prompt injection or obfuscation techniques were identified in the instructions or the code.

alibabacloud-pcap-analyzer