since-cutoff

Warn

Audited by Gen Agent Trust Hub on Nov 5, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill uses uvx to download and run the since-cutoff package from PyPI, specifically pinned to version 0.4.1. This constitutes the execution of a third-party dependency not maintained by a verified or well-known organization.\n- [COMMAND_EXECUTION]: Executes shell commands via the since-cutoff CLI for project scanning, synchronization, and testing. These operations have read access to the local project structure and source code.\n- [DATA_EXFILTRATION]: The skill makes network requests to pypi.org and the non-whitelisted domain models.dev to retrieve package releases and model training cutoff dates. The run command also sends package metadata and generated tasks to model providers to measure performance.\n- [INDIRECT_PROMPT_INJECTION]: The skill generates content for project documentation files (AGENTS.md or CLAUDE.md) based on external metadata and analyzed source code. This creates a surface where manipulated external data could theoretically inject instructions into the project's agent guidance files. This risk is managed by requiring explicit user confirmation before the skill writes any changes to disk.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 5, 2026, 02:54 PM