optimize-repo

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFEDYNAMIC_CONTEXT_INJECTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_CONTEXT_INJECTION]: The skill utilizes shell command execution at load time via the ! syntax to populate repository context information.\n
  • Findings: Commands such as git branch, git status, git tag, and jq are used to inspect the current environment metadata.\n
  • These commands are informational and do not incorporate user-supplied arguments, representing standard and safe context gathering.\n- [COMMAND_EXECUTION]: The skill instructs the agent to execute development scripts defined within the repository's configuration.\n
  • Findings: The instructions and the references/verify-ladder.md file specify running pnpm typecheck, pnpm lint, pnpm vitest, and pnpm turbo build.\n
  • These actions are standard for a repository optimization and verification workflow.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes repository source code and documentation which may contain untrusted data.\n
  • Ingestion points: The skill reads repository files, documentation, and configuration files (e.g., package.json) within the user-defined scope.\n
  • Boundary markers: There are no explicit markers or delimiters defined to isolate ingested file content from the agent's instructions during processing.\n
  • Capability inventory: The skill has permissions to read and modify local files and execute shell commands for build and test purposes.\n
  • Sanitization: Repository content is processed as-is without explicit sanitization or filtering.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 05:49 AM
Security Audit — agent-trust-hub — optimize-repo