otel-declarative-config
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches configuration schemas, documentation, and examples from the official OpenTelemetry GitHub organization (
open-telemetry/opentelemetry-configuration) and theopentelemetry.iodomain. These are well-known, trusted sources for the intended purpose of the skill. - [COMMAND_EXECUTION]: The skill uses the
gh(GitHub CLI) tool to perform read-only operations (listing and viewing releases) against a specific, hardcoded official repository (open-telemetry/opentelemetry-configuration). - [INDIRECT_PROMPT_INJECTION]: The skill is designed to handle untrusted data as a primary function. It contains explicit instructions for the AI to ignore embedded instructions in YAML comments or values, perform bounded inspection of inputs (checking size, depth, and node count), and use schema-only loaders that reject non-core YAML tags. This significantly mitigates the risk of indirect prompt injection from malicious configuration files.
- [REMOTE_CODE_EXECUTION]: While a test fixture (
evals/files/hostile-config.yaml) contains malicious patterns such as command substitution ($(curl ... | sh)) and unsafe YAML tags (!!python/object/apply), these are explicitly provided as negative test cases for the agent's evaluation. The main skill instructions provide robust mitigation strategies to identify and block such execution attempts.
Audit Metadata