skills/bklit/bklit-ui/pr-open/Gen Agent Trust Hub

pr-open

Pass

Audited by Gen Agent Trust Hub on Oct 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill automates a sequence of development commands, including git (status, add, commit, push), pnpm (lint, build), and gh (GitHub CLI) for pull request creation. These are standard operations for a development-focused agent skill.
  • [EXTERNAL_DOWNLOADS]: The workflow utilizes npx ultracite for linting and formatting. While npx downloads and executes packages from the registry, this is consistent with the vendor's stated development environment and is used to maintain code standards.
  • [INDIRECT_PROMPT_INJECTION]: The skill reads and processes untrusted content from the local environment via git diff to generate pull request summaries.
  • Ingestion points: Reads the output of git status, git diff, and git log in Step 1.
  • Boundary markers: The skill effectively uses shell HEREDOCs (cat <<'EOF') to isolate user-generated and repository-derived text from the shell execution environment, though it lacks semantic markers to prevent the LLM from interpreting instructions found within the code changes themselves.
  • Capability inventory: The agent possesses the capability to modify the local filesystem (git commit, pnpm lint:fix), access the network (git push, gh pr create), and execute build scripts (pnpm build).
  • Sanitization: There is no explicit sanitization or filtering of the code diff content before it is interpolated into the prompt for summarization.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 15, 2026, 10:37 PM