skills/cloudposse/atmos/atmos-hooks/Gen Agent Trust Hub

atmos-hooks

Pass

Audited by Gen Agent Trust Hub on Sep 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill documents the use of kind: command and kind: step (such as shell or container types), which allow the execution of binaries and shell scripts during lifecycle events like before.terraform.plan or after.terraform.apply.
  • [EXTERNAL_DOWNLOADS]: The instructions describe how Atmos resolves and installs required binaries specified in dependencies.tools to ensure they are available on the system path for hook execution.
  • [INDIRECT_PROMPT_INJECTION]: The hook system supports conditional execution using CEL expressions that process runtime facts such as stack and component names. 1. Ingestion points: Stack manifests and runtime metadata. 2. Boundary markers: The documentation does not specify explicit delimiters or sanitization for CEL expressions. 3. Capability inventory: The skill facilitates the execution of shell commands, containers, and infrastructure scanners. 4. Sanitization: Standard CEL evaluation is used without explicit mention of sanitization for user-provided configuration strings.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 26, 2026, 12:23 PM
Security Audit — agent-trust-hub — atmos-hooks