novoads-update

Pass

Audited by Gen Agent Trust Hub on Aug 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: Executes the local script './scripts/update.sh' and several 'git' subcommands (fetch, rev-parse, log, diff, rev-list) to perform status checks and repository updates.
  • [PROMPT_INJECTION]: Ingests and summarizes data from 'CHANGELOG.md' and git logs without sanitization or boundary markers, creating an indirect prompt injection surface. Ingestion points: CHANGELOG.md and git commit logs (SKILL.md). Capability inventory: shell execution, file writes, and network access via git.
  • [EXTERNAL_DOWNLOADS]: Performs network operations via 'git fetch' to retrieve update data from the origin remote.
  • [COMMAND_EXECUTION]: Modifies the platform's '.claude/settings.json' file to register a 'SessionStart' hook, allowing automated background updates based on user consent.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 13, 2026, 10:18 PM
Security Audit — agent-trust-hub — novoads-update