AI agent governance, security tooling, and mechanical enforcement.

rigscore v2.2.0

  • #rigscore
  • #security
  • #ai
  • #devtools
  • #open-source

Changes: v2.1.0 → v2.2.0

10 features and 9 fixes. 19 changes total.

Features

  • read Codex TOML and Goose YAML MCP surfaces
  • CLI strictness, output emitters, badge/SARIF/score-history, pin & verify-state hardening
  • BOM sniff, registry-derived scan surfaces, worktree skip, wider fixers + CI providers
  • fix vendor MCP/cred paths + add qwen sandbox, new clients, agents/workflows skillDirs
  • drive skill and command dirs from the client registry
  • add codex-guards least-privilege starter pack
  • make the semantic-judge command configurable
  • register OpenClaw and Antigravity for config-at-rest scanning
  • rename-safety parity + suppress/guards/fixture drift guards
  • machine-readable rigscore-facts.json emitter + self-gate

Fixes

  • cover the committed .codex/config.toml in the rug-pull pin
  • atomic state writes, platform-aware permission tests, script load fixes
  • emit POSIX separators in findings, SARIF and baseline keys
  • flush help output and load checks via file URLs
  • decouple HOME from project score, scan agent prompts + CLAUDE.local.md, tighten skill-files/git-hooks/settings
  • retire obsolete toolpage version-patch; stamp docs date directly
  • check out default branch for facts generator (tag lacks it)
  • warn on unknown/dangling flags; fail loud on a malformed –fail-under
  • honor scan opts, rescore suppress/ignore, harden –watch on Node 18

Install

1npx github:Back-Road-Creative/rigscore

No accounts, no telemetry, no network calls. MIT licensed.

github.com/Back-Road-Creative/rigscore

Try it → GitHub → All releases →
← Back to Releases