skills/coldteadotai/pr-lens/eli5/Gen Agent Trust Hub

eli5

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the user to install a companion skill from the same author using the command npx skills add coldteadotai/pr-lens.
  • [COMMAND_EXECUTION]: Executes git remote get-url origin to retrieve repository ownership and naming metadata required for the visual graph's provenance data.
  • [REMOTE_CODE_EXECUTION]: Uses npx @coldtea/pr-lens-cli@latest to run validation, rendering, and publishing commands. The CLI tool is a vendor-owned resource matching the skill author.
  • [DATA_EXFILTRATION]: The canvas push command uploads generated codebase architecture data to prlens.dev. This is the core functionality intended to create a shareable web-based walkthrough.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes codebase content, which is potentially untrusted, to generate explanations and diagrams.
  • Ingestion points: Reads local source files, directory structures, and git metadata from the project being analyzed.
  • Boundary markers: The output is strictly formatted into a predefined JSON schema (.pr-lens/graph.json).
  • Capability inventory: Employs shell command execution and network operations through the vendor CLI.
  • Sanitization: Implements strict vocabulary constraints (e.g., "plain words only", "3 words or fewer" for labels) which limits the complexity of text ingested from the codebase.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 06:08 AM
Security Audit — agent-trust-hub — eli5