Guards

No lazy placeholders: flags TODOs, '... rest of code', NotImplemented stubs and empty bodies an edit introduces, nudges Claude to finish them, and lists what is left.

One prompt, several jobs: when a prompt you type holds independent tasks (a list, numbered steps, "do A, then B, and also C"), offers to run them one by one as their own turns, stopping at the first that fails or asks you something.

Runs your project's checks after Claude edits, shows pass/fail above the prompt, and closes the loop with one-press (or automatic, bounded) fixes.

A little game layer: 32 achievements for how you use Claude Code, with lifetime progress and a /achievements pane.

Gob as a Tamagotchi-style companion who lives off your work: finished turns, green tests and commits grow him, and /pet lets you feed, play with, pet, nap and dress him. Free, and he never punishes you.

Opt-in: after a long turn, hear one spoken sentence about what Claude did.

Jev (TypeSafe's System One model) as Claude's instant co-processor: bulk filter/classify/rank/semantic-grep tools, risk notes under permission dialogs, a check on unverified 'it passes' claims, and optional per-prompt effort.

Opt-in instant approval for clearly harmless shell commands: a strict pre-filter, then Jev classifies the command; only read-only inspection at ≥ 97% with no network risk skips the permission dialog. Every approval is logged.

No narrating comments: every comment an edit adds is judged by Jev in a few hundred ms; 'Added X as requested' and friends go back to Claude.

Instant repo-convention checks on every edit: CLAUDE.md, AGENTS.md, style guides and config facts compiled into atomic rules, each edit judged against all of them by Jev in a few hundred ms.

Docs that went stale because of this turn's changes: changed names, flags, env vars and config keys are looked up in the docs, and Jev reads each section against the new code.

A definition of done Jev checks after every turn: per-criterion probabilities that demand evidence, a band above the prompt, optional auto-continue.

Notifications only when you are actually needed: after each turn Jev reads whether Claude asks you something, is stuck, or finished a long task; permission prompts ping too. Toast, chime, optional webhook.

A commit gate that reads the staged diff in a blink: Jev flags debug output, commented-out code, TODOs, machine paths, secrets and off-message changes before Claude's git commit runs.

"This helper already exists": new functions are compared against a git-grep symbol index, Jev judges the look-alikes in a few hundred ms, and Claude hears about duplicates.

Keeps changes on task: every edit is judged by Jev against your last request (and Claude's plan) in a few hundred ms; drive-by changes go back to Claude.

An instant security reviewer on every code edit: ~20 yes/no checks per edit answered by Jev in a few hundred ms, fed back to Claude.

Catches tests being weakened to make them pass: every test-file edit is judged by Jev in a few hundred ms against what you asked for.

Every failed Bash/PowerShell command gets an instant Jev diagnosis (15 failure classes) and a playbook in this project's terms, as a note Claude reads.

Huge Bash/PowerShell outputs cut to the lines that matter: Jev scores every 12-line chunk against your request, the full log is saved for Claude to open.

What Claude touched this session: reads, edits (+/− lines), searches per file, main vs subagent, with flags for blind edits and repeated reads.

Git at a glance and in reach: a branch/changes status line, a /git pane to diff, stage, revert and commit, and Conventional Commit messages written by haiku.

Project memory across sessions: every few turns haiku writes what was done, decided and left open to .claude/journal.md, and new conversations start with the latest entries.

Inline #tokens that attach live context to a prompt: #diff, #staged, #log:20, #tree, #todo, #errors, #file:path:10-40, #url:https://...