Skip to content

v0.2.1

Choose a tag to compare

@humanetresearch humanetresearch released this 10 Nov 09:34
· 4 commits to main since this release

What's new in v0.2.1

  • Separate advisory robustness_hints: Moved requests / urllib calls omitting timeout= out of behavioral_findings into robustness_hints (category: "robustness_hint"), so HTTP resilience hints no longer trigger ACTION_REQUIRED_STATIC_FINDINGS or dilute runtime asset-download precision (frozen rules commit 47b2ddc).
  • Fresh Phase 3 out-of-sample held-out benchmark: Evaluated frozen 47b2ddc rules on 10 previously unseen AI repositories (smolagents, crawl4ai, aider, open-webui, browser-use, markitdown, crewAI, openai-agents-python, servers, instructor) with real --network none sinkhole probes: 53 TP, 4 FP, 7 FN (93.0% precision, 88.3% recall).
  • Documentation & parser honesty fixes: Clarified built-in minimal TOML/YAML readers (stdlib only, no external parser), marked Phase 2 v0.2.0 re-run explicitly as non-out-of-sample, clarified v0.1.0 (NXDOMAIN) vs v0.2.0+ (sinkhole) timing differences, updated Python and Node --run copy, and deduplicated benchmark notes.