otel-declarative-config
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONREMOTE_CODE_EXECUTIONDYNAMIC_EXECUTIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external YAML configuration files which may contain untrusted instructions. It mitigates this risk through mandatory non-constructing parsing, rejection of custom YAML tags, and explicit instructions to ignore embedded directives in data.
- [REMOTE_CODE_EXECUTION]: A test fixture in
evals/files/hostile-config.yamlcontains a shell command injection pattern ($(curl ... | sh)) used to evaluate the agent's ability to detect and block executable content in data. - [DYNAMIC_EXECUTION]: The evaluation file
evals/files/hostile-config.yamlincludes an unsafe!!python/object/applyYAML tag intended to test the agent's robustness against dangerous object construction during parsing. - [CREDENTIALS_UNSAFE]: A fake authentication token is present in the
evals/files/hostile-config.yamltest fixture to verify the skill's capabilities for identifying and redacting secrets in diagnostics. - [PROMPT_INJECTION]: An 'ignore prior instructions' directive is included in a test fixture as an adversarial test case to ensure the agent maintains instruction-following integrity when processing external data.
Audit Metadata