workers-best-practices
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill fetches data from external sources (Cloudflare documentation) and processes local project files (e.g., wrangler configuration) to provide code reviews. This behavior creates a surface for indirect prompt injection where malicious instructions embedded in those sources could influence the agent's actions.
- Ingestion points: The skill fetches best practices from
https://developers.cloudflare.com/workers/best-practices/workers-best-practices/and reads project files includingwrangler.jsoncandnode_modules/wrangler/config-schema.json. - Boundary markers: No explicit boundary markers or instructions to ignore embedded directives were found for these input sources.
- Capability inventory: The skill utilizes shell commands including
mkdir,npm pack,tar,cat, andnpx(fortsc,eslint, andoxlint). - Sanitization: No evidence of sanitization or filtering of the ingested external content was identified.
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to manage type definitions, extract package archives, and run development tools for code validation.
- [EXTERNAL_DOWNLOADS]: The skill performs network requests to Cloudflare's official developer documentation site and downloads the
@cloudflare/workers-typespackage from the NPM registry to ensure type accuracy.
Audit Metadata