Skip to content
CAI
Software that uses CAICheck a score

SaarthurR/atticus-mac

51.8

Adequate · 16 November 2026

15.9k

lines of production code

Swift

with Python, JavaScript

1

measurement over time

CAI band scale
CAI lens gauges

What this system is

Behavioural changes

156 commits (21 fixes) modifying (repo-wide)

A change to existing behaviour in (repo-wide) — 156 commits (21 fixs), 13 files: “Fewer false error reports (a cancel or a "no" is not a crash), and Atticus recovers when its AI worker quits mid-answer”; “Atticus 1.1: reads first, then acts”; “Atticus: a voice assistant for your Mac”; and 153 more (refs \#1, \#21, \#27, \#4, \#25).

(repo-wide) · low confidence · unverified

91 commits (15 fixes) modifying ui

A change to existing behaviour in ui — 91 commits (15 fixs), 10 files: “Atticus 1.1: reads first, then acts”; “Atticus: a voice assistant for your Mac”; “Agents shelf: click to open, one orb per state; new and finished agents pop open, then fold back. Atticus remembers the agents you started”; and 88 more (refs \#27, \#4).

ui · medium confidence · unverified

Test coverage

110 commits adding/updating tests in tests

Test coverage in tests — 110 commits (19 fixs), 104 files: “Fewer false error reports (a cancel or a "no" is not a crash), and Atticus recovers when its AI worker quits mid-answer”; “Atticus 1.1: reads first, then acts”; “Agents shelf: click to open, one orb per state; new and finished agents pop open, then fold back. Atticus remembers the agents you started”; and 107 more (refs \#27, \#4).

tests · high confidence · unverified

Written by cca.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

How this codebase got here

Baseline

  • First survey — no prior run to compare against. CAI 51.8.

Lenses

  • Code Health 70.4
  • Architecture 99.7
  • Maturity 53.8
  • Readiness 50.8
  • Security 68.0
  • Accessibility 46.7
  • Performance 60.0

Changes since last survey

  • 170 commits — 147 feature/other, 23 fixes

By area

  • (root) — 158 commits
  • tests/onboarding.html — 2 commits
  • docs/icon.png — 1 commit
  • tests/init.py — 1 commit
  • tests/evals — 1 commit
  • tests/test_box_page.py — 1 commit
  • tests/test_fix_flow.py — 1 commit
  • tests/test_repo_fixes.py — 1 commit
  • tests/test_stream.py — 1 commit
  • ui/AppIcon.icns — 1 commit
  • ui/settings.html — 1 commit
  • ui/ui.html — 1 commit

Notable commits

  • fix: Agents start and finish clearly (sound, answer-first card, open their work from any ask), pointing with a guide line, live dictation wave, mini screen only when needed, faster and cheaper asks, safety and bug fixes
  • fix: Atticus 1.3.1: smart Clean up dictation, whiteboard explanations, Point at your screen button, routine and shortcut fixes
  • fix: Atticus 1.3: Dictate anywhere, Draw on screen, point at your screen, routines, named agents, Intelligence modes, Move to agent, calendar changes, local models and Pi, light mode, speed and reliability fixes
  • fix: Atticus 1.4.0: everything follows your display, new dictation pill, diagrams, learns on first success, fixes
  • fix: Bug sweep: drawings start right the first time, Insert survives closing the box, plain pasted text, tables and code in answers, one Try again, listening orb when you talk about a drawing
  • fix: Builds on macOS 26 SDKs again (fixes #21)
  • fix: Cleanup round: 60+ bug, safety and quality-of-life fixes; shared helpers; Route harder tasks to Sonnet setting
  • fix: Dictation cleans up properly: fillers like "like" and "you know" go, misheard tech words get fixed, long dictations get paragraphs. ⌃⌘V pastes your last dictation anywhere, and a dictation that couldn't go in shows a History button
  • fix: Fix Settings and read-aloud voices, and handle setup and cancellation safely
  • fix: Fix app focus, permission cards, and Claude failure handling
  • fix: Fix demo poster play button and keyboard playback
  • fix: Fix drawing captions, bottom timer status, and custom-server requests
  • fix: Fix: Settings no longer crashes when a custom server sends an odd model list
  • fix: Fix: a rare crash when a request sent the front app in an unexpected format
  • fix: Fix: an error after some answers since the last update
  • fix: Fix: answers did not show in the box
  • fix: Fix: the timer and current step show under your question again when you ask right after a dictation
  • fix: Fixes from live testing: the rest of a spoken sentence stays instant after a live move and unnamed screens mean the screen you're on; "When I say X, …" saves with no AI and shows Learned; emailing a person looks them up in Contacts (asks when unsure); CLI help picks the command path you need, capped at 8,000 chars; worker pipes can't crash cancellation
  • fix: Fixes: no crash for Pi and Devin users after an answer, no crash when the AI helper stopped early, arrows reach what they point at and stay 3 seconds, labels sit beside what they mark
  • fix: Revert "Fix Settings and read-aloud voices, and handle setup and cancellation safely"
  • …and 150 more

Written by cca.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

Survey your own repository

SaarthurR/atticus-mac was measured the same way every project in this corpus was: the same rubric, at a pinned commit, with the result published in full. Point a surveyor at a repository you know and see whether you agree with it.

About this page

  • The score is its most recent published measurement, taken on 16 November 2026 at a pinned commit. It is not a live figure and does not change until the project is measured again.
  • Measured at commit cee9340d5076d275ac90d321675f5ab0d89d3518 — the exact code this score is about.
  • Scored under rubric-2026.11.4 — the same rubric and the same method as every other entry in this index.
  • Measured by Canine Code Analyzer (cca.canine.dev) using cca measure-2026.11.84+preprod-d645ca9bbedd.