pulumi-debug-failed-operation

Pass

Audited by Gen Agent Trust Hub on Jul 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTIONNO_CODE
Full Analysis
  • [COMMAND_EXECUTION]: Employs the pulumi api subcommand to query operational data and utilizes standard command-line tools such as jq and sed for parsing JSON responses and cleaning up terminal formatting codes. These are standard operations for managing Pulumi stacks.
  • [DATA_EXFILTRATION]: Accesses operational metadata, stack names, and diagnostic logs from the Pulumi service. These network requests are directed to the vendor's official API endpoints using the local CLI configuration.
  • [PROMPT_INJECTION]: The skill processes diagnostic messages from external providers, which represents a potential surface for indirect prompt injection.
  • Ingestion points: Diagnostic and error logs retrieved from the Pulumi API as described in SKILL.md.
  • Boundary markers: No specific delimiters or warnings are used to isolate the diagnostic messages.
  • Capability inventory: File system modification for fixing code, shell command execution via the pulumi CLI, and stack state management.
  • Sanitization: No explicit sanitization or filtering of the message content is performed before interpretation.
  • [NO_CODE]: The skill consists entirely of markdown instructions and YAML configuration files, with no bundled executable scripts or binary files, which minimizes the risk of hidden malicious behavior.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 7, 2026, 04:56 PM
Security Audit — agent-trust-hub — pulumi-debug-failed-operation