absolute-upgrade

Pass

Audited by Gen Agent Trust Hub on Jun 30, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local commands for dependency management (npm, pip, go) and runs project-specific test, build, and lint scripts to verify upgrade safety. It also acknowledges the potential execution of third-party codemods during major version migrations.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted data from external sources during the dependency triage and migration phases.\n
  • Ingestion points: The skill reads ecosystem-specific manifests (package.json, requirements.txt, go.mod) and external documentation like changelogs and migration guides (SKILL.md).\n
  • Boundary markers: No specific delimiters or safety instructions are defined to separate untrusted external content from agent instructions.\n
  • Capability inventory: The agent has broad access to local shell execution for package management and testing across multiple languages (health-engine.md).\n
  • Sanitization: There is no evidence of sanitization or validation of the content retrieved from external changelogs or manifests.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 30, 2026, 03:07 AM
Security Audit — agent-trust-hub — absolute-upgrade