Skip to content

CHANGELOG.md

Latest commit

 

History

History
42 lines (33 loc) · 1.47 KB

CHANGELOG.md

File metadata and controls

42 lines (33 loc) · 1.47 KB
Raw

Changelog

All notable changes to SkillScan are documented here.

[0.4.0] - 2026-03-03

Added

  • Sandbox execution (skillscan test): run skills inside Docker Sandbox microVMs
  • Honeypot canary detection: plants fake credentials and flags exfiltration
  • Agent runtime selection (Claude Code, OpenClaw)
  • Test depth modes: quick (1), standard (3), deep (7) invocations
  • Evidence collection pipeline (network, filesystem, process events)
  • scan meta-command now chains audit + predict + test

[0.3.0] - 2026-03-03

Added

  • Predict mode (skillscan predict): LLM behavioral dry-run simulation
  • YAML-driven scenario system for behavioral simulations
  • Multi-provider support: Claude, OpenAI, Gemini, Ollama
  • Temporal and delayed-activation risk detection
  • scan meta-command combining audit + predict

[0.2.0] - 2026-03-03

Added

  • YAML rule engine with 59 rules across 7 categories
  • SARIF 2.1.0 output for GitHub Security tab integration
  • Rich terminal output with color-coded severity
  • skillscan install command with pre-install scanning
  • GitHub Action for CI/CD integration
  • 8 built-in analyzers: shell, macos, obfuscation, url, envvar, entitlement, typosquat, pattern
  • LLM-powered analysis (optional, via --llm flag)

[0.1.0] - 2026-03-03

Added

  • Initial project scaffolding
  • Core data models (ParsedSkill, Finding, ScanResult)
  • SKILL.md parser with frontmatter and code block extraction
  • CLI interface with audit command
  • Risk scoring engine