software-paas-hosting
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is entirely instructional and comparative, containing no executable scripts or system-modifying commands.
- [EXTERNAL_DOWNLOADS]: The skill references documentation and status pages for established cloud platforms (AWS, Vercel, Cloudflare, Fly.io). These references are used for fact-checking platform limits and are considered safe as they target well-known, trusted organizations.
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and verify data from external URLs listed in
data/sources.json. - Ingestion points: External documentation URLs defined in
data/sources.jsonand referenced inSKILL.md. - Boundary markers: The skill explicitly instructs the agent to "Verify before relying on" and "verify against data/sources.json and primary docs" before providing facts to the user.
- Capability inventory: No file-writing, subprocess execution, or network exfiltration capabilities are present in the skill's instruction set.
- Sanitization: No automated sanitization is performed, but the risk is mitigated by the advisory nature of the skill and the use of reputable vendor sources.
Audit Metadata