Skip to content

Order Samurai 2.1.4

Choose a tag to compare

View all tags
@Gemkai Gemkai released this 11 Oct 14:53
v2.1.4
1c4c565
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Order Samurai 2.1.4

Newer version available. v2.2.0 includes everything below, plus dashboard history that survives rebuilds and more accurate metrics. Install that one unless you need this exact version.

This release extends the prompt-injection guard beyond Claude Code and makes installs more dependable. If you're coming from 2.1.1, it also includes the 2.1.2 and 2.1.3 fixes listed at the end.

Protection for more coding agents

  • samurai install now installs the prompt-injection guard for Codex, Cursor and Gemini CLI as well as Claude Code. It detects which ones you have and prints one result line for each. Choose them yourself with --harness claude,codex,cursor,gemini or SAMURAI_HARNESS.
  • Install and doctor now name agents found on your machine that Order Samurai can't protect yet: Windsurf, Goose, Cline, OpenCode, Aider and Continue. That way a machine running Claude Code alongside one of them isn't mistaken for fully covered.
  • samurai install exits with an error when it finds no supported agent, instead of reporting success with nothing protected.
  • Each new agent has coverage limits, listed in full in the CHANGELOG. In short:
    • Codex: runs the guard only after you approve it in /hooks.
    • Cursor: doesn't cover file reads or Tab completions.
    • Gemini CLI: runs no hooks in folders you haven't marked as trusted.
    • Doctor: checks each one's setup but doesn't claim it's enforced.

More reliable installs

  • The web installer retries when the zip and its checksum don't match just after a release, instead of failing.
  • On Homebrew and other "externally managed" Pythons, install puts its dependencies in a private environment, ~/.samurai/venv.
  • If a hook registry already runs the guard and scrubber, install doesn't register them a second time.
  • Re-installing leaves unchanged hooks alone. Uninstall removes only what Order Samurai wrote, and no longer reports success when it couldn't remove a hook.
  • Config files are written safely: changes are atomic, permissions are kept, and an unreadable ~/.claude/settings.json is never replaced.

Clearer failures

  • The daily digest prints a warning before falling back to Mail.app, naming the error but never the key.
  • Doctor's hook-health rows say why a hook's criticality is unknown when the hook registry can't be loaded.
  • execution/verify_falsifiability.py reports a crashing check as an error and finishes the rest, instead of stopping with a traceback.

Also new

  • One support address: support@agentica-llc.biz. It's printed at the end of samurai doctor and after a successful install.

From 2.1.2 and 2.1.3

  • 2.1.3: secret scrubbing works on a standard install. In 2.1.1 the scrubber depended on a file a standard install doesn't include, so it scrubbed nothing. samurai doctor now runs each hook for real and fails if one errors or hangs.
  • 2.1.2: dashboard. Previous and next controls for categorised reflex stacks, and the selected card stays put when other alerts clear.

Full details are in CHANGELOG.md.

Assets 4
Loading