The Agent Skills Directory

[SAFE]: The skill serves a legitimate development purpose: tracking updates to AI runtime specifications and flagging 'drift' in a local installer file.
[SAFE]: Network operations are directed at fetching documentation from established technology platforms and official project domains. No data exfiltration or unauthorized network communication was identified.
[SAFE]: File system interaction is restricted to reading its own reference files and a specific local installer script (bin/install.js) using standard search and read tools.
[SAFE]: Indirect Prompt Injection analysis (Vulnerability Surface):
Ingestion points: Documentation URLs listed in SKILL.md and references/baseline.md processed by WebFetch.
Boundary markers: None present.
Capability inventory: WebFetch, WebSearch, Read, Glob, Grep, and Bash (specifically for grep on local files).
Sanitization: None present.
Assessment: The ingestion surface is required for the tool's core research function; no malicious instructions or exploits were found in the static content.

gspdev-runtime-compat