convex-bundle-size

Pass

Audited by Gen Agent Trust Hub on Nov 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses npx to execute various utility commands for analysis and deployment.
  • Evidence includes calls to npx -y @convex-dev/bundle-analyzer summary, npx -y @convex-dev/bundle-analyzer insights --json, npx convex codegen, and npx convex dev --once. These are standard tools provided by the vendor to assist in debugging and managing the deployment infrastructure.
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to download and run packages from the Convex vendor ecosystem at runtime.
  • It utilizes npx -y @convex-dev/bundle-analyzer, which fetches the latest version of the analyzer tool if it is not already cached locally. As this originates from the official vendor scope (@convex-dev), it is considered a legitimate resource for the skill's stated purpose.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from the local project environment, which could theoretically contain malicious content intended to influence the agent's behavior.
  • Ingestion points: Reads output from @convex-dev/bundle-analyzer (which analyzes local project files), package.json, and convex.json from the user's workspace.
  • Boundary markers: The instructions do not explicitly define boundary markers for the JSON output of the analyzer, but the agent is instructed to process the data locally for the purpose of bundle optimization.
  • Capability inventory: The skill has the capability to modify project configuration files (package.json, convex.json) and source code to apply bundle-shrinking fixes.
  • Sanitization: None described. The risk is minimized as the data is sourced from the user's own project and official vendor tooling.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 20, 2026, 08:14 PM