dart-pub-upgrade

Pass

Audited by Gen Agent Trust Hub on Jul 31, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill relies on executing a provided Dart script (scripts/bin/dart_pub_upgrade.dart) which spawns several shell processes including git status/checkout/add/commit, dart pub/fix/analyze, flutter pub, fvm, and melos. It also instructs the agent to use the GitHub CLI (gh) to create pull requests.
  • [EXTERNAL_DOWNLOADS]: The included Dart script performs network requests to the official package registry (pub.dev) to fetch metadata and download source archives (tarballs). It extracts CHANGELOG.md files from these archives in-memory to generate diffs.
  • [PROMPT_INJECTION]: The skill contains a surface for indirect prompt injection (Category 8) because it fetches untrusted content from the web and asks the AI to process it.
  • Ingestion points: Untrusted text from CHANGELOG.md files on pub.dev is saved to changelog_diffs.json and subsequently read by the AI for summarization.
  • Boundary markers: Absent. There are no explicit instructions or delimiters used to warn the AI about potential instructions embedded within the changelog content.
  • Capability inventory: The agent has significant capabilities including filesystem access, shell command execution (via the skill script), and the ability to create Pull Requests using the gh CLI.
  • Sanitization: No sanitization or filtering is performed on the changelog text before it is presented to the AI.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 31, 2026, 05:20 AM
Security Audit — agent-trust-hub — dart-pub-upgrade