dependency-upgrader

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources including package manifests, lockfiles, and vulnerability advisory databases which could potentially contain malicious instructions.\n
  • Ingestion points: Dependency manifests (package.json, pom.xml, build.gradle), lockfiles (package-lock.json, yarn.lock), and output from auditing tools (npm audit, gh api, osv-scanner). Found in SKILL.md and referenced playbook files.\n
  • Boundary markers: The skill does not define technical delimiters for ingested data, relying instead on natural language instructions for the agent to distinguish tool output from core instructions.\n
  • Capability inventory: Shell command execution for dependency management and testing; file system write access for manifest updates; network access via CLI tools for registry and advisory lookups.\n
  • Sanitization: Instructions emphasize manual verification, reviewing diffs before execution, using ignore-scripts where supported, and comparing findings against a baseline audit.\n- [COMMAND_EXECUTION]: Executes standard package manager and build system commands (npm, pnpm, yarn, bun, gradle, maven) and security CLI tools (gh, osv-scanner) to resolve dependency trees and run security scans.\n- [EXTERNAL_DOWNLOADS]: Fetches metadata, release notes, and security advisories from well-known services such as GitHub, NPM, Maven Central, and the OSV database. These actions are consistent with the skill's primary purpose and use official repositories.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 09:45 PM
Security Audit — agent-trust-hub — dependency-upgrader