dotagents-qa
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses a Node.js orchestration script (
scripts/qa-example.mjs) to execute various CLI commands includinggit,claude,codex,opencode, andpi. These executions are limited to testing the dotagents integration and are performed within an isolated Docker container. - [EXTERNAL_DOWNLOADS]: The Dockerfile and documentation provide instructions to download and install official agent CLI tools and the dotagents package from the NPM registry. These resources originate from trusted organizations (Anthropic, OpenAI, Sentry, etc.) and are used for their intended primary purpose in a development/QA context.
- [DYNAMIC_EXECUTION]: The QA script utilizes dynamic
import()to load local repository modules (list.js,scope.js) and usesexecFileSyncto run shell commands based on the local environment. It also dynamically generates configuration files (TOML, JSON) to create isolated test fixtures. - [INDIRECT_PROMPT_INJECTION]: The skill handles untrusted data by ingesting plugin manifests and skill definitions during testing. The risk is mitigated by the use of a Docker sandbox boundary and explicit instructions to avoid persistent credential storage in the output directories.
Audit Metadata