Prompt file imported from cwmat/frame-ripper (.codex/prompts/opsx-ext-codify-bulk.md). Copyright stays with the author.

/opsx-ext-codify-bulk

Bulk-codify multiple existing capabilities into baseline OpenSpec specs in one coordinated session.

Argument (optional): comma-separated list of capability names, OR a directory to focus discovery on (e.g., src/utils/). If omitted, the workflow discovers boundaries across the whole src/ tree and proposes them for user approval.

This is a one-time brownfield onboarding tool — for adopting OpenSpec on a stable project all at once. For incremental adoption (one capability when you happen to be working in that area), use /opsx-ext-codify instead.

The workflow has a mandatory planning phase before any spec is written: it discovers candidate capability boundaries, shows the list, and waits for the user to confirm/edit/add/remove before looping. That checkpoint is what separates this from the "auto-discover and pray" pattern that produces shallow specs.

Read .github/skills/openspec-ext-codify-bulk/SKILL.md and follow its workflow exactly. Pay particular attention to:

• The clean working tree preflight
• The high-churn check (bulk codifying actively-changing code produces specs that decay fast)
• The Phase 2 approval gate — never skip it
• The quality floor — halt if 2 consecutive specs come out with 0 requirements rather than continuing to produce noise

Setup note: this prompt file lives at .codex/prompts/opsx-ext-codify-bulk.md in the repo. To enable it as a Codex CLI slash command, symlink or copy it into ~/.codex/prompts/.