aws-observability
Pass
Audited by Gen Agent Trust Hub on Jul 24, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill's documentation guides users to download AWS Distro for OpenTelemetry (ADOT) agents and installation scripts from official AWS-related GitHub repositories (e.g.,
github.com/aws-observability). These are well-known, trusted sources for AWS telemetry tooling. - [COMMAND_EXECUTION]: The skill utilizes several local Python scripts (e.g.,
di_instrumentation.py,di_snapshots.py) to interface with AWS Application Signals and CloudWatch Logs APIs. These scripts are invoked via the shell tool and are designed with security-focused patterns, such as using an internal if-else dispatcher to avoid dynamic attribute access (getattr) and reading JSON arguments from files/stdin to prevent shell injection. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes CloudWatch Logs, which are considered untrusted data originating from running applications. An attacker could potentially embed malicious instructions in application logs to influence the agent. The skill mitigates this by providing explicit operating contract rules, such as a 'Sample-first' rule and a strict prohibition against using runtime data to drive breakpoint placement.
Audit Metadata