smoke-test
Pass
Audited by Gen Agent Trust Hub on Apr 25, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill documentation indicates that it performs automated fixes by downloading software from the Bun project (bun.sh) and the NPM registry (zod@4).
- [REMOTE_CODE_EXECUTION]: The auto-fix for the Bun runtime involves downloading and executing content from bun.sh at runtime.
- [COMMAND_EXECUTION]: The skill uses the 'exec' tool to run local scripts (scripts/smoke-test.sh) and dynamically executes any executable shell scripts found within the '~/.gbrain/smoke-tests.d/' directory.
- [CREDENTIALS_UNSAFE]: The skill is configured to check for the presence of embedding API keys and reads the 'GBRAIN_DATABASE_URL' environment variable. Diagnostic output, which may include details about these configurations, is logged to '/tmp/gbrain-smoke-test.log'.
- [PROMPT_INJECTION]: The skill provides an extensibility mechanism that serves as an attack surface for indirect prompt injection. Ingestion points: Shell scripts in the '~/.gbrain/smoke-tests.d/' directory. Boundary markers: None; the skill executes any file matching the glob pattern. Capability inventory: Uses the 'exec' tool and shell execution. Sanitization: None; script contents are executed without verification.
Audit Metadata