cloudflare-r2

Pass

Audited by Gen Agent Trust Hub on Sep 4, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references official Cloudflare documentation and utilizes well-known dependencies for R2 interactions, including aws4fetch, wrangler, and hono. It also mentions the official @aws-sdk/client-s3 package. These references target trusted organizations and established service providers, serving legitimate functional purposes without escalating security risk.
  • [INDIRECT_PROMPT_INJECTION]: The skill provides templates for object storage which inherently ingest untrusted data from client requests (e.g., via c.req.arrayBuffer() in templates/r2-simple-upload.ts).
  • Ingestion points: templates/r2-simple-upload.ts (c.req.arrayBuffer), references/common-patterns.md (c.req.formData)
  • Boundary markers: Not implemented within the data storage templates
  • Capability inventory: env.MY_BUCKET.put, env.MY_BUCKET.delete, and env.MY_BUCKET.list capabilities are provided to the agent
  • Sanitization: Standard data storage handling is used without content-level sanitization, which is typical for object storage functionality.
  • [SAFE]: The skill explicitly warns against exposing sensitive credentials like R2_ACCESS_KEY_ID in client-side code and provides clear instructions for using environment variables and secrets. All documented patterns align with standard Cloudflare development practices.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 4, 2026, 06:55 AM
Security Audit — agent-trust-hub — cloudflare-r2