darc

# DARc — Deep Adversarial Review CLI Iterative adversarial review: alternating Claude self-review and OpenAI GPT-5.6 cross-model review, with synthesis at the end of every round and after all rounds complete. ## Routing Parse `$ARGUMENTS`: | Argument | Action | |----------|--------| | (empty) or `<target-file>` | Run full review with default 3 rounds | | `<number>` | Run `<number>` rounds. 0 = unbounded with cap (ask at 20). 00 = no cap, runs until goal | | `help`, `--help`, `-h` | Show help | | `doctor`, `--doctor`, `check` | Run doctor | | `config` | Show current configuration (provider, model, env vars) | | `config provider <codex\|openai\|claude>` | Set adversarial model provider | | `config model <id>` | Set model (validated against current provider) | | `version`, `--version`, `-v` | Print version | **Argument parsing:** If `$ARGUMENTS` starts with a digit, that's the round count. `00` is a special case — it means no cap (unbounded until goal met), distinct from `0` (cap at 20 with human continue). All other args are treated as the target file path. ## Privacy Warning **OpenAI rounds upload the target file content to OpenAI's API.** Do not use this skill on files containing secrets, credentials, PII, or proprietary data you cannot share with a third-party API. Content is transmitted over HTTPS and processed by OpenAI under their data usage policy: https://openai.com/policies/api-data-usage-policies ## Target File The target file is the artifact to review. If no...