opty

Appearance

Getting Started ​

opty is a local-first CLI for AI agent configuration files. This page is generated from the main repository README so the install path and command surface stay in sync.

Install ​

bash
npm install -g optymd

Or use directly:

bash
npx optymd scan .
npx optymd scan CLAUDE.md

Core workflow ​

opty init claude   → scaffold config for your platform (claude | openclaw)
opty scan .        → detect files, find issues, show report
opty score .       → calculate quality score (0-100) and grade (A-F)
opty fix --dry-run → preview what will be fixed
opty fix .         → apply safe fixes (backup created automatically)
opty restore       → undo last fix from backup
opty verify .      → compare repo rules with local git, workflow, and GitHub evidence
opty verify . --json → emit machine-readable verify output
opty verify . --fail-on-violations → exit with code 2 when any verify rule is violated

Every fix creates a timestamped backup in .opty/backups/. Restore any snapshot with opty restore --id <timestamp>. For CI usage, opty verify . --fail-on-violations keeps the normal report output and returns exit code 2 when at least one rule is VIOLATED.

First report ​

$ opty scan .

  opty scan - claude
  ________________________________________________________
  5 files  |  Score: 72/B  |  1 critical  4 warnings

  TRUNCATED FILES (content silently cut off)
  x skills/bureau-dev/SKILL.md  2969/2000 (33% lost) trim ~969 tokens

  DEAD REFERENCES (4 broken paths)
  ! `skill.md` (referenced 2x)
  ! `tailwind.config.js`
  ! `heartbeat.md`

  NOTABLE FILES
  Best:
  A CLAUDE.md                    92  no issues
  B skills/drizzle-ops/SKILL.md  74  no issues
  D skills/bureau-dev/SKILL.md   47  2 minor issues

  BY TYPE
  agent      1 file   avg ##########---- 92
  skill      3 files  avg ######------ 53
  config     1 file   avg ########---- 80

  FIX PRIORITY
  1. Trim 1+ files exceeding token limits
  2. Remove 4 dead file/script references
  3. Add scope boundaries to 2 skills

  Run opty fix to auto-fix safe issues (secrets, verbose phrases, duplicates)